i

3 1293 01046 8316

LIBRARY
Michigan State

in minimumMn um m
Unlverslty

 

 

 

 

This is to certify that the

dissertation entitled

The Role of Dimba Land and Small Scale Irrigation in
Smallholder Farmers' Food Security in Malawi:
An Application of Safety First Chance-Constrained
Target Motad Mathematical Programming

presented by

Naomi Aretha Ngwira

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Agricultural Economics

M4 72/35;

Major professor

 

 

Date November 11, 1994

 

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

 

PLACE ll REI’UMI Boxwmwofltbchoekomflomyumd.
TOAVOID FINES Munonorbdorodatoduo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An Animativo Action/Ema! Opponmily Inflation
Wanna-m

 

 

 

.‘

,

W ""44

A /-

,1?"";
f

' " ‘ i
"« "1.5"",‘h‘4l' "11?";3 (Ml
. f I

"‘¢u4r+

,1; ‘

"mu

‘.Q'

I d

.g.p

THE RC
5m
AV A]

 

 

 

THE ROLE OF DIMBA LAND AND SMALL SCALE IRRIGATION IN
SMALLHOLDER FARMERS’ FOOD SECURITY IN MALAWI:
AN APPLICATION OF SAFETY FIRST CHANCE-CONSTRAINED
TARGET MOTAD MATHEMATICAL PROGRAMMING

By

NAOMI ARETHA NGWIRA

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

erIEEI
§

5.

 

 

ABSTRACT

THE ROLE OF DIMBA LAND AND SMALL SCALE IRRIGATION IN
SMALLHOLDER FARMERS’ FOOD SECURITY IN MALAWI:
AN APPLICATION OF SAFETY FIRST CHANCE-CONSTRAINED
TARGET MOTAD MATHEMATICAL PROGRAMMING

By

Naomi Aretha N gwira

The potential of intensive dimba (flood recession) land cultivation and small scale
irrigation to improve smallholder farmers’ food security is investigated. The objectives are
to: 1) estimate the current contribution of dimba crops and potential contribution of irrigated
crops to household food security; 2) identify technically feasible crop/irrigation technology
packages for increasing crop production; and 3) select those that can contribute most to food
security.

Smallholder farmers make up 80% of Malawi’s population. The majority are food
insecure owing to low productivity agriculture. Yet adoption of intensive cultivation has been
constrained by limited access to inputs and yield risk due to erratic rainfall. Cultivation on
dimba (flood recession) and irrigated land may offer a solution.

A survey of smallholder farmers was conducted in Chiradzulu North Extension
Planning Area of Blantyre Agricultural Development Division. The data collected are used to
implement a Target MOTAD programming model to select optimal crop activities. It uses
Telser’s safety first criterion to model risk avoidance behavior. Current, intensive and
irrigated crop production practices are analyzed for small (0.7 hectares), medium (> 0.7 to
1.49 hectares) and large farms (> 1.49 hectares). Sensitivity analysis is done for alternative
output prices, costs and crop yields.

Results show that dimba land and irrigation can contribute from 30% to 80% of total
income depending on farm size and crop production practices. With current cropping
practices, only large farms can achieve minimum cereal consumption requirements in 75 % of

cropping seasons. Even these larger farms need intensive dimba cultivation or irrigation to

generate incomes that meet subsistence requirements in 75 % of cropping seasons. Small
farms can produce cereal consumption requirements in 75 % of the cropping seasons only
when irrigation is used for cereals; they can achieve subsistence income requirements only
when horticultural crops are irrigated.

When farmers use an income safety-first strategy, stable incomes from intensive
dimba cultivation or irrigation of horticultural crops can make it optimal to adopt maize
fertilization or hybrids on rainfed land. Intensive dimba cultivation and small scale irrigation
require credit and extension programs to encourage crop diversification. Small farms with no
dimba can benefit from programs for cereal irrigation, as well as horticultural crops, where

markets for the latter exist.

Copyright by
NAOMI ARETHA NGWIRA
1994

"Since the rains have been insufficient, the people will now concentrate on cultivating on
moist areas along the dambos and rivers." Louis Chimango, M. P., Minister of Finance,
Daily Times, March 16, 1992.

DEDICATION
To my dear husband Chimwemwe
who provided the support I needed to pursue my doctoral studies.

vi

ACKNOWLEDGEMENTS

I wish to acknowledge with gratitude the contributions made by members of my dissertation
and guidance committee, Dr. James Shaffer, Dr. Michael Weber, Dr. Ann Ferguson and Dr.
Scott Swinton. A special word of thanks goes to Dr. Swinton for ably advising and

encouraging me through all the stages of preparing the dissertation.

vii

CHAP

 

Glam

TABLE OF CONTENTS

LIST OF TABLES ............................................. xii
LIST OF APPENDIX TABLES ..................................... xvi
LIST OF FIGURES ............................................ xix
CHAPTER 1: INTRODUCTION ..................................... 1
1.1 The Concept of Household Food Security ........................ 1

1.2 Food Security Among Smallholder Farmers in Malawi ................ 2

1.3 Objectives of the Study .................................... 4

1.4 Need for the Study ...................................... 4

1.5 Related Research ........................................ 6

1.6 Organization of the Study ................................. 10
CHAPTER 2: AGRICULTURE AND IRRIGATION POLICIES IN MALAWI ...... 11
2.1 Macro-Economic Performance .............................. 11

2.2 Agricultural Development Policy ............................ 14

2.2.1 The Estate Sector of Agriculture ....................... 14

2.2.2 The Smallholder Sector of Agriculture ................... 16

2.3 Irrigation Potential in Malawi ............................... 19

2.3.1 Hydrological Resources ............................ 19

2.3.2 Irrigation Policies ................................ 21

2.4 Implementation of Irrigation Projects .......................... 25

2.5 Summary ........................................... 28
CHAPTER 3: ANALYTICAL METHOD .............................. 29
3.1 Smallholder Farmer Objectives .............................. 29

3.2 Decision Making Under Uncertainty ........................... 31

3.2.1. Risk and Uncertainty ............................. 31

3.2.2. Models of Risk Analysis in Agriculture .................. 32

3.2.1.1 Direct Expected Utility Maximization ............. 35

3.2.1.2 Mean-Variance Analysis ...................... 39

3.2.1.3 Mean-Semivariance (Lower Partial Moments) ..... A. . . 40

3.2.1.4 Stochastic Dominance ....................... 41

3.2.1.5 Minimization of Total Absolute Deviations (MOTAD) . . 42

3.2.1.6 Target Minimization of Total Absolute Deviations ..... 44

3.2.1.7 The Safety First Approach .................... 45

3.3 Telser’s Safety First Probablistic Criterion in Target MOTAD .......... 47

3.4 Conclusion .......................................... 49

viii

CHAPTER 4: DATA COLLECTION ................................. 50

4.1 Selection of Study Area .................................. 50

4.2 Sampling ............................................ 50

4.3 The Questionnaires ..................................... 51

4.4 Preparation for the Survey ................................. 52

4.5 The Survey .......................................... 53

4.6 Secondary Data Collection ................................. 53

4.7 Summary ........................................... 54

CHAPTER 5: CHARACTERISTICS OF SURVEY HOUSEHOLDS ............. 55

5.1. Household Characteristics ................................. 55

5.1.1 Social and Demographic Characteristics .................. 55

5.1.2 Size and Distribution of Land Holdings .................. 55

5.2 Farm Characteristics .................................... 58

5.2.1 Cropping Patterns ................................ 58

5.2.2 Estimating Yields ................................ 61

5.2.3 Family Labor Available and Use ....................... 61

5.2.4 Production Expenses .............................. 68

5.2.4.1. Hired Labor Expenses ...................... 68

5.2.4.2 Fertilizer Expenses ......................... 68

5.2.4.3 Pesticides Expenses ........................ 73

5.3. Income Generating Activities and Off—Own Farm Employment .......... 73

5.4 Maize Production and Food Security .......................... 76

5.4.1 Maize Harvests ................................. 76

5.4.2 Food Security and Income Strategies .................... 78

5.5 Farmers’ Willingness to Use Irrigation ......................... 79

5.6 Summary ........................................... 80
CHAPTER 6: ENTERPRISE BUDGETS AND FORMULATION OF THE TARGET

MOTAD MODEL ........................................ 82

6.1 Enterprise Budgets ...................................... 82

6.1.1. Estimating Yields ............................... 83

6.1.1.1 Current Crop Enterprises ..................... 83

6.1.1.2 Intensive Crop Enterprises .................... 84

6.1.1.3 Irrigated Crop Enterprises .................... 85

6.1.2 Valuing Yields and Inputs ........................... 86

6.1.3 Variable Costs .................................. 87

6.1.3.1 Seeds ................................. 87

6.1.3.2 Fertilizer and Chemicals Applied to Crops .......... 87

6.1.3.3 Repairs and Maintenance ..................... 88

6.1.4 Labor Use and Costs .............................. 88

6.1.4.1 Family Labor ......................... -. .. 88

6.1.4.2 Hired Labor ............................. 89

6.1.4.3 Marketing Costs ........................... 89

6.1.5. Irrigation Costs ................................ 90

6.2 Formulation of the Mathematical Programming Model ............... 91

6.2.1 The Objective Function ............................ 92

6.2.2 Activities in the Model ............................. 92

ix

6.2.2.1 Crop Production Activities .................... 93

6.2.2.2 Labor Hiring Activities ...................... 93
6.2.2.3 Labor Selling Activities ..................... 93
6.2.2.4 Crop Selling Activities ....................... 94
6.2.2.5 Deviations and Chance Counters ................ 94
6.2.3 Constraints .................................... 94
6.2.3.1 Family Labor ............................ 95
6.2.3.2 Land .................................. 95
6.2.3.3. Accounting Rows ........................ 106
6.2.3.4 Deviation and Chance Constraint Rows ........... 106
6 3 Summary .......................................... 107
CHAPTER 7: MODEL I - CURRENT CROPPING PRACTICES .............. 108
7.1 Base Solutions ....................................... 111
7.1.1 Maize Production Approach and the Non-risk Constrained Solutions 111
7.1.1.1 Activities in the Solutions ................... 111
7.1.1.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises ............................. 1 16
7.1.2 Safety First Solutions ............................. 119
7.1.2.1 Activities in the Solutions .................... 119
7.1.2.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises ............................. 120
7.1.2.3 Contribution of Dimba Crops to Income .......... 121
7.1.2.4 Maximum Feasible Maize Production ............ 121
7.2 Sensitivity Analysis .................................... 121
7.2.1 The Non-risk Constrained Solution .................... 124
7.2.1.1 Activities in Solutions ..................... 124
7.2.1.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises ............................. 124
7.2.1.3 Contribution of Dimba Crops to Income ........... 125
7.2.2 Safety First Solutions ............................. 126
7.2.2.1 Activities in Solutions at the 0.25 Probability Limit . . . 126
7.2.2.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises ............................. 126
7.2.2.3 The Contribution of Dimba Crops to Income ........ 126
7.3 Lessons from Model I .................................. 126
7.4 Validating the Model ................................... 128

CHAPTER 8: MODEL II - CURRENT AND INTENSIVE CROPPING PRACTICES . 130
8.1.1 Non-Risk Constrained Solutions and the Maize Production Approach] 30

8.1.1.1 Activities in Solutions ...................... 130
8.1.1.2 Shadow Prices, Stability of Solutions and Competitiveness of

Enterprises ........................... -. . 135

8.1.1.3 Contribution of Dimba Crops to Income ........... 136

8.1.1.4 Maximum Feasible Maize Production ............ 137

8.1.2 Safety First Solutions ............................. 138

8.1.2.1 Activities in the Solutions .................... 138
8.1.2.2. Shadow Prices, Stability of Solutions and Competitiveness of

Enterprises ............................. 138

8.1.2.3 Contribution of Dimba Crops to Total Income ...... 139

8.1.2.4 The Maximum Feasible Maize Production ......... 139
8.3 Lessons from Model 11 .................................. 139
CHAPTER 9: MODEL 111 - CURRENT, INTENSIVE AND IRRIGATED
CROPPIN G PRACTICES ................................... 141
9.1 Base Solutions ....................................... 141
9.1.1 Model III(a): Irrigated Vegetables .................... 141
9.1.1.1. Activities in Solutions ..................... 141
9.1.1.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises ............................. 142
9.1.1.3 Contribution of Irrigated Crops to Total Income ...... 147
9.1.1.4 Maximum Feasible Maize Production ............ 148
9.1.2 Sensitivity Analysis for Model III(a) ................... 148
9.1.2.1 Sensitivity to Changes in Cost of Irrigation ......... 148
9.1.2.2 Sensitivity to Changes in Prices of Output (With Sprinkler
Irrigation) .............................. 149
9.1.2.3 Sensitivity to Changes in Wages ................ 150
9.1.3 Model III(b): Irrigation of Cereals .................... 150
9.1.3.1. Activities in Solution ...................... 151
9.1.3.2. Shadow Prices, Price Stability and Competitiveness of
Solutions .............................. 151
9.2 Lessons from Model 111 ................................. 152
CHAPTER 10: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ...... 154
10.1 Summary of Objectives and Research Methodology ............... 154
10.2 Summary of Farm Survey Results .......................... 155
10.3 Summary of Mathematical Programming Results ................. 156
10.3.1 Cropping Activities ............................. 156
10.3.2. Expected Income and Contribution of Dimba Crops to Total Incomd57
10.3.2.1. Current Cropping Practices ................. 157
10.3.2.2 Intensive Cropping Practices ................. 159
10.3.2.3 Irrigated Crop Practices .................... 159
10.3.4. Shadow Prices of Resources and Constraints ............. 162
10.3.5. Maximum Feasible Maize Production ................. 162
10.4. Lessons from Analysis of Programming Results ................. 162
10.5 Recommendations for Dimba Cultivation and Irrigation ............. 164
10.6 Recommendations for Future Research ....................... 165
APPENDIX A .............................................. 167
APPENDIX B .............................................. 208
BIBLIOGRAPHY ............................................ 223

xi

Table
lrriga:

Table
A Sim

Table
Socio-
Table
82: .1

Table
Mesa 1

Table
M1351 ]

Table
Yields
Table
Yields
Table

Table
Famil}

Table
Comp,

Table
[Sc 0‘

I

Table .
Hired
Table ,

 

LIST OF TABLES

Table 2.1

Principal Domestic Export Commodities 1992 ............................ 12
Table 2.2

Irrigation Development (Hectares) ................................... 23
Table 3.1

A Simple Linear Programming Table ................................. 34
Table 5.1

Socio-Economic Characteristics of Households by Size of Landholding ............ 56
Table 5.2

Size (Hectares) and Distribution of Land Holdings of Households ................ 57
Table 5.3

Most Dominant Crop Mixtures ..................................... 59
Table 5.4

Most Important Enterprises by Area (Hectares) Covered ...................... 60
Table 5.5

Yields of Rainfed Crops Enterprises (kg/ha) ............................. 63
Table 5.6

Yields of Dimba Crop Enterprises (kg/ha) .............................. 64
Table 5.7

Comparing Local Variety Maize Yields Mean kg per ha) with Other Reports ........ 65
Table 5.8

Family Labor Allocated to Crop Enterprises (Median Person-hours) .............. 66
Table 5.9

Comparison of Labor Allocation to Maize/Beans Crop Mixture (Person-hours) ....... 67
Table 5.10

Use of Hired Labor (Person-hours) by Enterprise by Crop Production Activity ....... 70
Table 5.11

Hired Labor (Person-hours) by Crop Enterprise ........................... 71
Table 5.12

Use of Fertilizer kg/ha by Crop Enterprise and Size of Farm ................... 72

xii

Table 5.13

Main Occupation of Household Heads ................................. 75
Table 5.14

Proportion of Farm Households Engaged in Non-Farm Work by Farm Size ......... 75
Table 5.15

Mean Rainfed Maize Harvests (Kilograms) by Size of Farm in 1989 - 1992 ......... 77
Table 5.16

Mean Harvests of Fertilized and Unfertilized Maize (Kilograms) ................ 77
by Size of Farm in 1991/92

Table 5.17

Food Security and Income Strategies by Farm Size (Percent of Sample) ............ 79
Table 5.18

Constraints to Producing More Crops ................................. 80
Table 6.1

Yields of Irrigated Crop Enterprises .................................. 86
Table 6.2

Costs of Various Types of Irrigation .................................. 91
Table 6.3

Model I - Current Rainfed and Dimba Crop Enterprises ...................... 97
Table 6.4

Model 11 - Intensive Crop Enterprises ................................. 98
Table 6.5

Model III(a) - Irrigated Horticultural Crops Using Sprinklers - Income Approach ...... 99
Table 6.6

Model III(b) - Irrigated Cereals Using Gravity/Furrows - Maize Production Approach . . 100
Table 6.7

Model III(b) - Irrigated Cereals Using Gravity/Furrows - Income Approach ........ 101
Table 6.8

Labor Hiring Activities ......................................... 102
Table 6.9

Labor Sale Activities ........................................... 103
Table 6.10

Crop Sale Activities ........................................... 104

xiii

Ill

Ch;

Tab
Mal
Pan

Tabl
Mod
Part1

Table
Mode
Contr

Table
Model
by All

Table '
Model

Table '7
Model I

Table 7
Madel 1
and Off-

Table 3
“Niel 11
Panza] 5

Table 3
“Ede! I]
and Am
L951 of

Table 8.
M1161 II

Table 3,.
”0161 ll
91 Off.;

”Cale! II

Table 6.11

Chance Activities ............................................. 105
Table 7.1

Model 1 - Safety First Solutions Using Telser’s Probablistic Criterion and Atwood’s Lower
Partial Moments At Probability Limits 0.25, 0.33 and 0.4OSmall Farms ........... 112
Table 7.2

Model I - Safety First Solutions with Telser’s Probablistic Criterion and Atwoods Lower
Partial Moments At the 0.25 Probability Limit Level of Activities in Solution . . . . 113-114
Table 7.3

Model I - Base Scenario

Contribution of Dimba Land, Rainfed Land and Off-farm IEAs to Income ......... 117
Table 7.4

Model I - Maximum Feasible Maize Production (kg) Goals

by Allowable Probability of Failure .................................. 118
Table 7.5

Model I - Shadow Prices of Resource and Constraints ...................... 122
Table 7.6

Model I - Sensitivity Analysis Results - Small Farms ....................... 123
Table 7.7

Model I - Contribution of Dimba, and Rainfed Land Crops

and Off-Farm IEAs to Total Income Under Three Scenarios Small Farms .......... 125
Table 8.1

Model 11 - Safety First Solutions Using Telser’s Probablistic Criterion and Atwood’s Lower
Partial Moments At Probability Limits 0.25, 0.33 and 0.40 - Small Farms ......... 132
Table 8.2

Model II - Safety First Solutions with Telser’s Probablistic Criterion
and Atwoods Lower Partial Moments At the 0.25 Probability Limit
Level of Activities in Solutions .................................... 133

Table 8.3
Model 11 - Shadow Prices of Resource and Constraints ...................... 136

Table 8.4
Model I] - Contribution of Dimba and Rainfed Land
and Off—Farm IEAs to Total Income in Non-risk Constrained Solution ............ 137

Table 8.5
Model 11 - Maximum Feasible Maize Production Goals by Allowable Probability of Failurel38

xiv

Table 9.1
Model III(a) - Safety First Solutions Using Telser’s Probablistic Criterion
and Atwood’s Lower Partial Moments Using the Income Approach

At Probability Limits 0.25, 0.33 and 0.40 - Small Farms .................

Table 9.2
Model III(A) - Safety First Solutions with Telser’s Probablistic Criterion
and Atwood’s Lower Partial Moments Using the Income Approach

At the 0.25 Probability Limit Level of Activities in Solutions ..............

Table 9.3

Safety First Premium of rainfed, Dimba and Irrigated Land in Models 1, II and III . . . .

Table 9.4
Model III(a) - Contribution of Irrigated and Rainfed Land

and Off-Farm IEAs to Total Income ..............................

Table 9.5

Model 111 - Returns Under Various Types of Irrigation for Small Farms ........

Table 10.1
Comparison of Expected Returns, Maximum Feasible Goals and Maize Production

Models 1, II and III .........................................

XV

143

144-145

146

149

161

Table
Enter;

Table .
Enterp

Table .1
Enterpr

Table .11
Enterpr

Table A
Enterpn

Table At
Enterpm
l

Table A6
Enterprg

Table A»

Emitters

 

 

 

Table Ag
Emerpris

Table .116
Emmi-WIS:

Table A5
Emeran:

Table A6

‘9
VILJPYIS:

Tamc .
c1 ‘*5
E

'39
NEWTTM

Table As
Emery,“

Emma

\

LIST OF APPENDIX TABLES

Table A6.]

Enterprise Budget for 1 Hectare of Current Non-Fertilized Local Maize ........... 168
Table A6.2

Enterprise Budget for 1 Hectare of Current Fertilized Local Maize .............. 169
Table A6.3

Enterprise Budget for 1 Hectare of Intensive Local Maize .................... 170
Table A6.4

Enterprise Budget for 1 Hectare of Hybrid Maize ......................... 171
Table A6.5

Enterprise Budget for 1 Hectare of Current Non-Fertilized Maize/Beans ........... 172
Table A6.6

Enterprise Budget for 1 Hectare of Fertilized Local Maize/Beans ............... 173
Table A6.7

Enterprise Budget for 1 Hectare of Intensive Fertilized Maize/Beans ............. 174
Table A6.8

Enterprise Budget for l Hectare of Hybrid Maize/Beans Irrigated Using Gravity/Furrows 175
Table A6.9

Enterprise Budget for 1 Hectare of Hybrid Maize/Beans Irrigated Using Pump/Sprinklers 176
Table A6.10

Enterprise Budget for 1 Hectare of Current Non-Fertilized Local Maize/Beans/Pigeon Peas 177
Table A6.11

Enterprise Budget for 1 Hectare of Current Fertilized Local Maize/Beans/Pigeon Peas . . 178
Table A6. 12

Enterprise Budget for 1 Hectare of Intensive Local Maize/Beans/Pigeon Peas ....... 179
Table A6. 13

Enterprise Budget for 1 Hectare of Hybrid Maize/Beans/Pigeon Peas ............. 180
Table A6. 14

Enterprise Budget for l Hectare of Current Fertilized Local Maize/Beans/Groundnuts . . 181
Table A6.15

Enterprise Budget for 1 Hectare of Intensive Local Maize/Beans/Groundnuts ........ 182

xvi

Table
Enter

Table
Eaten

Table
Eater;

T3516 A"
E11183“,
Table A.
Emtmn.

‘ ;
Table Ag
Emcrpns

Tiblc A6
Enter-ms
[sing Pu
Table .ae
EmCEP-‘is

Title M,
5112311an

Table A6.16

Enterprise Budget for l Hectare of Rice Irrigated Using Gravity/Furrows ......... 183
Table A6.17
Enterprise Budget for 1 Hectare of Rice Irrigated Using Pump/Sprinklers .......... 184
Table A6. 18
Enterprise Budget for l Hectare of Wheat Irrigated Using Gravity/Furrows ......... 185
Table A6.19
Enterprise Budget for 1 Hectare of Wheat Irrigated Using Pump/Sprinklers ......... 186
Table A6.20
Enterprise Budget for 1 Hectare of Current Relay Rainfed Field Peas ............ 187
Table A6.21
Enterprise Budget for l Hectare of Current Dimba Field Peas ................. 188
Table A6.22
Enterprise Budget for 1 Hectare of Sugarcane ........................... 189
Table A6.23
Enterprise Budget for 1 Hectare of Current Non-Fertilized Cabbage ............. 190
Table A6.24
Enterprise Budget for 1 Hectare of Intensive Cabbage ...................... 191
Table A6.25

Enterprise Budget for 1 Hectare of Cabbage Irrigated Using Pump/Sprinklers-Season 2 . 192

Table A6.26
Enterprise Budget for 1 Hectare of Cabbage Irrigated
Using Pump/Sprinklers-Seasons 1 and 3 ............................... 193

Table A6.27
Enterprise Budget for l Hectare of Current Non-Fertilized Tomato .............. 194

Table A6.28
Enterprise Budget for 1 Hectare of Intensive Tomato ....................... 195

Table A6.29
Enterprise Budget for 1 Hectare of Irrigated Tomato Using Pump/Sprinklers-Season 2 .' . 196

Table A630

Enterprise Budget for 1 Hectare of Irrigated Tomato
Using Pump/Sprinklers-Seasons 1 and 3 ............................... 197

xvii

Table
Enter
Seasa

Table
Emer
Seam

Table
Enter

Table
Enter

 

 

Table 1'
Enter .

Seasc~

Table 11

Enter
Seaso

Tible 9
Enter

Table

Enter

 

Table A6.31
Enterprise Budget for l Hectare of Current Chinese Cabbage ................. 198

Table A6.32 .
Enterprise Budget for 1 Hectare of Intensive Chinese Cabbage ................. 199

Table A6.33
Enterprise Budget for 1 Hectare of Irrigated Chinese Cabbage Using Pump/Sprinklers-
Season 2 .................................................. 200

Table A6.34
Enterprise Budget for 1 Hectare of Irrigated Chinese Cabbage Using Pump/Sprinklers-
Seasons 1 and 3 .............................................. 201

Table A6.35
Enterprise Budget for 1 Hectare of Non-fertilized Dimba Maize ................ 202

Table A6.36
Enterprise Budget for 1 Hectare of Qirrent Non-fertilized Dimba Maize/Pumpkin Leaves 203

Table A6.37
Enterprise Budget for 1 Hectare of Irrigated Green Beans Using Pump/Sprinklers-
Season 2 .................................................. 204

Table A6.38
Enterprise Budget for 1 Hectare of Irrigated Green Beans Using Pump/Sprinklers-
Seasons 1 and 3 .............................................. 205

Table A6.39
Enterprise Budget for 1 Hectare of Irrigated Onions Using Pump/Sprinklers-Season 2 . . 206

Table A6.40

Enterprise Budget for 1 Hectare of Pumpkin Leaves
Irrigated Using Pump/Sprinklers-AII Seasons ............................ 207

xviii

.I'
('Y
‘1
1

lf‘ 1’:

”Hr"
Final:
5-

Pretty.

 

 

LIST OF FIGURES

Figure 1: Map of Malawi Showing Agricultural Development Divisions and the Study Area 18
Figure 2: Map of Malawi Showing Existing Irrigation Projects Area .............. 22

Figure 3: Illustration of the Concepts of Risk Aversion, Certainty Equivalence, and Risk
Premium ................................................... 37

xix

farm;

smell

1.1 T

 

 

 

CHAPTER 1: INTRODUCTION

The central focus of this study is the problem of food insecurity among smallholder
farming households in Malawi. The goal is to investigate the potential of dimbal land and

small scale irrigation to improve the food security position of these households.

1.1 The Concept of Household Food Security

In general, household food security can be defined as the ability of a household to
provide its members access to a timely, reliable, and nutritionally adequate food supply
(World Bank, 1986 p 216). For Malawian farmers, the concept of food security is defined by
two factors. Staple food is maize. Except in a few ecological zonesnthe northern lake shore,
the southern and northern tip of the country—maize has no competition from secondary staples
such as cassava, rice, and sorghum.

Peters and Herrera (1989) have aptly recorded the sentiments of smallholder
Malawian farmers with respect to maize:

”Chimanga ndi moyo," say the farmers in Zomba Southzz "Maize is our life."

Nothing captures so forcibly the goal of all households—to produce all the maize they

could possibly want. In this area maize is equated with food...the first criterion of

wealth is to have enough maize to eat; to have land is to grow maize. More than an
income generating activity in the dry language of analysts, cultivating maize is a way
of life... to cultivate without cultivating maize was an impossible thought.

Maize is grown once a year under very uncertain rainfall conditions. This, added to

the situation of incomplete and imperfect markets, makes farmers desire to produce as much

 

' Dimba land is flood recession land along rivers and in depressions. It is cultivated after
the rainy season.

2 Zomba South is an EPA contiguous with Chiradzulu North.

1

 

hear 1.-

hOUS ,3.
Else :5.

ill
we? [es
hf‘idi} S_t
f ., ,
ma Ca:
shoe n“
We e31
Conan: 1

 

2
as possible of their maize consumption requirements. Again, Peters (1991) has succinctly
recorded this motivation:

The people’s wish to produce as much of their own maize as possible is related to
both their assessment of their ability to pay the higher prices in deficit period
(December-January) as measured against their needs for cash immediately after maize
harvest; the insecure set of income opportunities available to them, which are more
restricted for the poor; and the fluctuation in supply at the local ADMARC selling
centers, as well as costs in time and cash, involved in buying maize. For these
reasons, as well as for taste preferences, the more of their food supplies people are
able to produce for themselves, the less risk they face (of increased prices or of short
supplies).

Thus for the Malawian case, household food insecurity is primarily the failure of a
household to produce enough maize to meet its annual consumption requirements. But not all
households can produce their maize requirements from their land holdings. Because of very
small land holdings, a large majority depend on off-farm income earning activities (IEAs) or
sale of horticultural crops produced on dimba land. Thus, the operational definition of
household food security for this study is the ability of households to produce maize and/or

generate enough cash to meet any the shortfall in maize production.

1.2 Food Security Among Smallholder Farmers in Malawi

Smallholder farming households make up 80% of Malawi’s population. For policy
analysis and planning, the Ministry of Agriculture groups these households into three
categories based on the size of their land holdings: those with less than 0.7 hectare of land
holdings, those with 0.7 to 1.49 hectares, and those with more. This grouping is related to
food security issues and to the kinds of income and employment support programs that
farmers can benefit from. The group with less than 0.7 hectares represents those households
who cannot meet their subsistence needs from their land alone with present crop production
technology. This category constitutes 36.7% of all smallholder farming households in Malawi
and 57.4% in Blantyre Agricultural Development Division (ADD) where this study was
conducted.

1.3 Objectives of the Study

The basic questions the study will address are:

a) What is the current contribution of dimba production to food security and income in

Chiraduzlu Extension Planning Area (EPA)?

b) Can intensive dimba cultivation and small scale irrigation be used to improve the food

security situation?

The specific objectives are:

1) to understand the current pattern of land use of both dimba and rainfed land
for agricultural production in Chiradzulu North Extension Planning Area
(EPA);

2) to compile enterprise budgets for farm activities congruent with land and time
use as detemtined above;

3) to estimate the contribution of dimba crops to household farm income and
food security;

4) to estimate the possible contribution of irrigated crops to household income
and food security;

5) to identify technically feasible crop/irrigation technology packages that can be
used to increase production of crops;

6) to select crop-technology packages that can contribute most to improving food
security;

7) to make policy and research recommendations based on the findings of the
study to appropriate institutions.

1.4 Need for the Study

Alternative ways of improving smallholder food production and security are needed

for the following reasons:

farmers
“ere 1":
hells b
I
“lit less
aim-3:3
“Tier: 1T1:

 

 

5

1) At present, many farmers have too little land to produce adequate food given

current technology and farming systems.

2) There are limited employment opportunities outside agriculture, and irrigated

agriculture may absorb underemployed labor.

3) Rainfall is unreliable, and adequate means for insuring against crop loss and

failure are not available.

Recent studies indicate that maize research has released semi-flint hybrid maize
varieties, e.g. MH18, with characteristics desired by farmers (Smale et al 1993). Wide-
spread adoption is still to occur. In the past, the available non-flint hybrids did not meet the
farmers’ storage and processing preferences. Farmers grew local maize varieties. Yields
were very low due to limited access to inputs (Kydd 1989, Smale 1991). The result was
yields below annual consumption requirements of households.

Employment opportunities outside agriculture are very limited. For those households
with less than 0.7 hectares of land the government has suggested a policy of finding
alternative employment (Malawi Government 1990). However, there is no suggestion of
where they can get work or how to generate the jobs. In 1987, the population census
enumerated 81.3% of the labor force as employed in the agriculture sector. Farmers who do
not work on their own farms work on other farmers’ gardens or in the estate sector. The
estate sector employed 9% of the labor force in 1990 (Mkandawire et a1 1990). The
conditions of employment in this sector are exploitative, not engendering long term household
food security. Intensification of smallholder agriculture is thus a more plausible option given
the economic base of the country.

In the past three cropping seasons (1990/91, 1991/92, and 1992/93), drought has led
to serious food shortages among as many as 90% of farming households in most of Central
and Southern parts of the country. Coupled with the organizational and financial uncertainties

of managing national grain reserves, the drought has put the majority of rural farms at risk of

lzlra:
“Perle:
Scale in

HO“BVQ

Optlorts i

   
  

”ngahon

sufflCICn:

 

 

 

 

6

starvation. The remarks quoted earlier from the Minister of Finance indicate that Malawian
farmers cannot rely on rainfed cultivation.

The nature of drought in Malawi and the existence of large water bodies make
irrigation an attractive albeit pecuniarily more costly alternative technology for agricultural
production and improving food security. Irrigation can serve both a) to alleviate the risk of
drought between one season and the next, and b) to smooth out intra—season fluctuation of
water supply to plants. It can also permit higher productivity cultivation practices, with a
direct impact on the volume of output and farm incomes. In this sense, irrigation is not just a
risk reduction strategy. It also has a major impact on output via its complementarily with
double cropping, increased fertilizer use, and improved seeds. In the past, neither dimba
cultivation nor expanded formal irrigation has been given due attention. The recent
experience with drought in the last three years has renewed interest in irrigation. A large
scale irrigation project is being implemented for the irrigation of the Lower Shire Valley.
However, intensive dimba cultivation and small scale irrigation may be the only feasible
options in other parts of Malawi.

There is inadequate information on how to plan and organize for small scale
irrigation. A pivotal question is whether irrigation with different kinds of techniques would
sufficiently improve food security to justify the requisite investments. This study proposes to
address this information gap by determining the crop mixtures that farmers can profitably
grow, given that their overriding objective is to maximize the probability of producing enough

maize for home consumption and/or earn income to buy those food requirements.

1.5 Related Research
Decision making concerning irrigation planning can be phased into three: The first
one is how much irrigation land to develop, given water resources and other uses of land; the

second level addresses the issue of what crops to grow in a given season; the third level deals

we
(I

 

7

with how to optimally allocate water to crops that are planted (Duddley 1971). This study
belongs to level two.

No in-depth farm level financial studies have been done on small scale irrigation in
Malawi. There is also a paucity of literature on irrigation in sub-Saharan Africa in general.
The little literature available is concentrated on the Gezira Scheme in Sudan and developments
along the Nile. Most of the irrigation projects existing in sub-Saharan Africa are traditional
or small scale self-help schemes run by farmers on which governments did not have to spend
large sums of money or carry out careful financial assessments. It is the analyses of these
large scale projects that are mostly published in international journals and media (Barnett
1994). What follows in this section is a review of financial studies of small scale irrigation
that were done in other parts of the world.

The closest studies in objectives and methods of this study are those conducted by
Maji and Heady (1978), Islam (1973), Preibprom (1982) and Laki (1992). Maji and Heady
dealt with the issue of how to allocate water to six irrigation blocks of a large project with
command area of 530,528 acres in Mayurikshi, India. They used a chance constrained linear
programming (LP) model to develop an optimal cropping pattern and a reservoir management
policy for the project. The Optimal plan allowed the probability of drought or floods to be
less than 10%. This was necessary to attract farmers who have low risk bearing capacity.

Islam analyzed the economic and financial returns to low-lift pump irrigation in
Bangladesh. He used an LP model to determine the impact of three different rental rates and
command area (area irrigated by one pump) as measured by net returns to fixed farm
resources, optimal cropping patterns, marginal value products, and amount of inputs used.
Economic returns were calculated using internal rate of return, net present value and
benefit/cost ratio. The results indicated that there would be substantial specialization of crops
under irrigation conditions, unless the yield of some crops increased substantially. Sensitivity
to yields and prices was very high. He concluded that the government of Bangladesh could

expand irrigation to all areas with suitable hydrological resources. Subsidies could be

in
q”;

{CIT

3.73.3

 

 

8

significantly reduced without adversely affecting farmers’ incentives. A package combining
rental charges, credit, and supply of equipment and inputs was necessary to expand command
area.

Priebprom (1982) approached the problem from a labor use point of view. He
appraised the alternative uses of farm, non-farm and off-farm work, and their impact on
production, employment and income of rural households in the Khon Kaen province of
Thailand. A poly-period LP model was used to represent farm households with three
different farm size groups for both rainfed and irrigated area. Major non-farm and off-farm
employment was included in the model to test the complementarity and competitiveness of
farm, non-farm enterprises and off-farm employment. He found that for rainfed farms, those
with small holdings derived most of their income from off-farm work. Both rainfed and
irrigated farmers could combine farm, non-farm enterprises and off-farm work to achieve
maximum net income.

Laki (1992) analyzed the impact of changes in resource use, enterprise mix on farm
income on a typical Gezira tenancy, and assessed the impact of government policy on
productivity. A static LP model was used to maximize gross margins, subject to minimum
sorghum consumption requirements of the tenant households. Sensitivity analysis was done to
analyze the impact of different technologies, policies and resource levels on cropping pattern,
farm productivity and income. The results showed that 1) labor, capital and water were
constraining during peak periods; 2) elimination of mandated cropping pattern increased
income; 3) levels of income achieved were sensitive to prices, yield levels and irrigation water
supply. The second finding is relevant to Malawi where farmers on rice settlement schemes
have been discouraged partly by coercive production regulations.

Faki (1982) used an LP model to examine the financial and managerial aspects of
irrigation water in the Gezira Scheme. Running his model at various capacities he concluded

that water was in short supply in October and November. Returns to water were high.

9

Cotton was the most profitable crop followed by sorghum and groundnuts, and wheat had
negative returns. He also found a cropping intensity of 60% below the planned 75%.

Hazelwood and Livingstone (1978) conducted studies in the irrigated Usangu Plains of
Tanzania. Using an LP model, they concluded that large and small scale farms are not
alternatives in the bid to maximize paddy production. In a separate study in 1982, they used
an LP to determine the demand for irrigation water, and also discussed issues surrounding
supply of water.

Ndiame (1985) used a production function econometric analysis to evaluate large and
small irrigation perimeters in the Flueve region of Senegal for their contribution to that
country’s objective of food security. He found that large scale perimeters led to more per
capita availability of food, even though small scale perimeters have higher yields. He
recommended more economical strategies of food security based on simultaneous development
of irrigation with rainfed and flood recession agriculture. This finding is relevant to Malawi:
flood recession (dimba) cultivation, and small scale capital extensive irrigation may be
complementary to large scale capital intensive irrigation.

Diallo (1980) did a comparative financial and economic analysis of capital and labor
intensive rice perimeters in Senegal River Valley. He found that large perimeters are less
efficient than small perimeters. Both small and large perimeters have production cost above
the total of purchase price, insurance and freight costs of rice from Asian countries to
Senegal. Domestic resource cost ratios were 125% and 233% for rice produced on small and
large perimeters respectively. Reasons given were cheap rice quality that is imported from
Asia, and inefficient irrigation policies followed by the country’s irrigation board.

The studies reviewed show that irrigation can be financially feasible under various
arrangements and has potential in improving incomes and meeting food requirements of-
fanners. This may be the case even when cropping intensity or command areas are lower

than optimum.

1.6 (

imgazi
this stL
outline
charaer:
6 coma]
progra'r
In ,fp'

resorting?

 

 

10

1.6 Organization of the Study

The study is organized as follows. Chapter 2 provides a synopsis of agriculture and
irrigation policies in Malawi. The rationale for the choice of methods and techniques used in
this study is presented in Chapter 3. The planning and implementation of the farm survey are
outlined in Chapter 4. Following in Chapter 5 is a presentation of the major descriptive
characteristics of the survey households, especially those connected to food security. Chapter
6 contains documentation on the compilation of enterprise budgets, and the formulation of the
programming model used. Chapters 7 through 9 present and discuss the results of the model.
In Chapter 10 is a summary of the findings and conclusions. It also contains

recommendations to appropriate agencies and for future research.

RESCUE
plllhceg
lad Gaffe
PW

‘1} a ;

 

813w ,.

mfg! <

 

CHAPTER 2: AGRICULTURE AND IRRIGATION POLICIES IN MALAWI

This chapter provides a synopsis of agricultural and irrigation potential and policies in
Malawi. It furnishes a macro—economic overview and illustrates the need for small scale
irrigation. The aim is to facilitate the understanding of research objectives and design, and

also to provide a context for assessing the findings of the study.

2.1 Macro-Economic Performance

Malawi got independence in 1964 from Britain, in one of the worst national economic
conditions and political legacy for a people whose leader was ambitious to lead them to
prosperity. The colonial government had strategic legislation concerning land and important
economic resources, the purposes of which was to entrench colonial hegemony (Nankumba
1981, Chipeta 1992). The country has no exploitable mineral resources and so it was treated
as a labor reserve for the South African mines and also Zimbabwe and Zambia, which
together with Malawi were part of the British Colonial federation.

The country’s economic development strategy is based on rapid increase in
agricultural productivity to boost export crop production and ensure adequate food supply. It
produces its own basic foodstuffs, and exports tobacco, tea, sugar, groundnut, rice, cotton
and coffee. Tobacco contributes 74% of export earnings. (See Table 2.1.)

The way that the agricultural sector has been organized to implement the policy is
purely a political matter. This is discussed in the following section. However, before that an
overview of the macro-economy and its performance is provided.

Between 1964 and 1979, the economy maintained a strong upward trend: GDP
growth rates averaged 5 %. The country was acclaimed for some the best performance .
amongst small landlocked countries and by the World Bank and the International Monetary
Fund as one of the examples of adopting their current policy prescriptions for least developed

countries, and doing well. Others feel that those who held out Malawi as a virtuous case

11

Tcagha
enem
llarch .
due to

(Malta

 

 

12

Table 2.1
Principal Domestic Export Commodities 1992

 

Agricultural crops Value(MK' million) Proportion(%)
Tobacco 1029.8 74.2
Tea 106.1 7.6
Sugar 97.8 7.0
Rice 2.0 0.1
Cotton 16.6 1.2
Pulses 7.0 2.0
Coffee 27.0 0.5
Other Exports 100.5 7.2
Total Domestic Exports 1387.4 100.0

 

tended to exaggerate the strengths while ignoring the weaknesses (Kydd and Christiansen
1981, Lele 1990). An important motif in their argument was that the pattern of growth
chosen was not optimal, especially when distributional issues are considered.

These misgivings were actually born out. Between 1979 and 1982, GDP growth
reached a standstill. Most of the immediate causes of reduction in economic activity were
external: rapid deterioration in Malawi’s terms of trade, a second rise in oil prices, a drought
which compelled maize imports, closure of Malawi’s main transport link to outside markets
due to the war in Mozambique, and historically high interests rates in international markets
(Malawi Government, series of Economic Reports).

A After implementing structural adjustment programs with lending from the IMF and
World Bank the economy was resuscitated, although the performance of key macro-economic
variables has been mixed. During 1983-1985 GDP grew at 4% from below zero in 1981.
Between 1985 and 1987 there was another decline; the economy picked up in 1988. Inflation
reached an all—time high of 31.4% in 1988 but fell to under 10% in 1989 due to availability of

goods under the import and industrial liberalization schemes that are part of structural

econ
point
better
has It

Come.
deem;
agncul

Product

13

adjustment. Poor performance in this period has been attributed to contractionary macro-
economic policies, which wrung growth prospects from the economy. Some recent analyses
point to poor entrepreneurship and management styles, but it is difficult to conceive what
better management would have achieved without resources (Meyers 1990). Besides, Malawi
has had very strong economic institutions and probably the least corrupt African civil service.

Under the structural adjustment, the poor have paid a heavy price (Harvey 1987,
Cornea et a1 1988, Kandoole 1989). They have suffered because social services were initially
de—emphasized due to the inflationary effect of currency devaluation and the way that
agricultural marketing liberalization increased uncertainty in food markets. Although
producer prices were increased as part of structural adjustment policies, the majority of
smallholder farmers have failed to respond to the incentive due to small land holdings and/or
no access to inputs (Lele 1990). Also, most are net purchasers of food, not sellers.

Despite these programs and policy changes, the performance of the economy is far
from favorable. The export base is still narrow, and the economy is still highly vulnerable to
exogenous and natural shocks. In 1992, ”real GDP declined by 7.9% compared to a growth
rate of 7.8% in 1991, mainly due to short-fall in small-scale agricultural production arising
from the drought” ([Malawi Government l993(c)] p4). The public sector still runs large
deficits. Between 1991 to 1993 the deficit in the government budget increased from MK328m
to MK848m (6.48% to 9.36% of total GDP). Investment as a ratio of GDP declined in the
same period. The balance of payments, which was in surplus in the 1990-1992, plunged to
MK505.5m in 1993. These economic data combined with high population growth (3.3%),
lead to a forecast of no per capita GDP growth between 1993 to 1996 ([Malawi Government
l993(c)] p21). An important lesson from the analysis is how drought can seriously impact the

macro-economy. These effects may be dampened with more extensive use of irrigation. -

22

export
allies 2
created

of mere
211]

reward '0
and cred:
and pofttz
preterenti;
Re
smallness
an ids-am
mlonb'ot

,,

“CC to

The
“0 e
0the.
[O in
com:

TOP

IS the 0m} f5

 

 

14

2.2 Agricultural Development Policy

Emanating from a mistrust of the capacity of smallholder farmers to grow and handle
export crops, and from a need to use the profits from estate production to reward political
, allies and create a support base, a dichotomous estate and smallholder agricultural sector was
created. The ultimate effect of this has been to deprive the rural poor of their potential source

of increased income through preferential land and marketing policies.

2.2.1 The Estate Sector of Agriculture

Until recently, the estate sector was the domain of pe0ple the government wanted to
reward or appease for political reasons. They have had the privilege of easy access to land
and credit. Commercial banks have been made to give agricultural loans to top civil servants
and politicians. These asset entitlements were complemented by the subsidization of inputs,
preferential marketing procedures and a supply of cheap labor.

Resources have been funnelled to the estate sector, while conspicuously slighting the
smallholder sector. Although government planning documents emphasized the need to ensure
an adequate food supply to save foreign exchange, there was no real effort to equip the
majority of the smallholder farmers with the capacity to produce food. With particular
reference to maize there was this explicit policy:

The increased production for maize should be a result of higher yields per acre, and
no encouragement should be given to growing maize on land at present used to grow
other crops. It is important that this be understood by all concerned as any tendency
to increase the proportion of cultivated land planted with maize would be directly
contrary to the nation’s long term objectives (Malawi Government 1979) p 16.

To produce more maize per unit of land, farmers need more fertilizer which at present

is the only feasible way of increasing production. But the allocations of fertilizer credit to

15

smallholder farmers have not increased at the same rate as credit to estate agriculture, despite
the poor credit record in the estate sector3 (Kydd and Christiansen 1982).

Estate and smallholder agricultural together now contribute nearly 40% to gross
domestic product (GDP) and provide over 90% of total domestic exports. However, since
1964 the estate sector has expanded considerably, such that smallholder agricultural output as
the percentage of total agriculture output fell from 93 % to 80% between 1965 and 1980 and is
now at 72%. In 1970 only 23.5 % of the tobacco area was under estate agriculture, but this
rose to 47% in 1985. Productivity has also been higher in the estate sector for export crops
like tobacco and tea. In the case of tobacco, yields in the smallholder sector are only half in
the estate sector. The yield and area factor have led to a situation now where 73% of tobacco
is produced in the estate sector compared to only 40% in 1970 (Malawi, Government Series,
Economic Reports 1970-1993).

The estate sector accounts for 80% of export earnings, mostly from tobacco, but
also from tea, coffee and sugar. It employs about 10% of the labor force. Recent studies of
the sector indicate that there has been a structural change in the sector with more ownership
by graduated smallholder farmers (Mkandawire et a1 1990). These smallholder farmers have
graduated by consolidating land with relatives and registering it as leasehold to enable them to
get quotas to sell high value burley tobacco to international buyers. This is certainly an
important policy of relevant finding.

However, the conclusion that leads one to question the long held position that
smallholder farmers have not benefited from development of export-led agriculture is

lopsided. It is certainly the case that the incomes of those who participate in estate sector

 

3 Fertilizer use intensity averages 154 leg/ha in the survey households compared to an
average of 328 kg/Ha in the estate sector, where most of the fertilizer is ’leaked’ from
ADMARC subsidized fertilizer intended for smallholder farmers (Mkandawire et a1 1990).

IT}:

pre

("I
”"3

lane
grad
Err:

from

2.2.2

as ba
"M the
SOLE-gm l
PTOJeCts
{Tillers 1
[he more
Schemes"
Ft
aPPTOaCh‘
pro”dad t

31":de and 1

l6
improves. However, any farmer who can consolidate or buy 30 hectares of land is not the
typical smallholder farmer discussed in policy issues in Malawi. It is the tenants who work in
these estates that fit into this category. Concerning them, the authors question "if under
prevailing price and cost conditions, tenancy arrangements can alleviate poverty on any
significant scale" (Mkandawire et a] 1990 p xiv). All estates account for only 20% of total
land. Besides, even if the graduated smallholders constitute half of the estate owners, the
graduated smallholder farmers are a minuscule proportion of smallholder farming families.
Evidence still supports the position that the majority of smallholder farmers have not benefited

from the development strategy based on large-scale agriculture.

2.2.2 The Smallholder Sector of Agriculture

Immediately after independence, the approach to smallholder agricultural development
was based on the transformation approach. But between 1965 and 1969, the government felt
thatthe efforts to develop agriculture without concerted programs would not achieve much. It
sought to concentrate on promising regions, "through the establishment of high productivity
projects while maintaining the policy of gradual improvement for the mass smallholder
framers through general extension and marketing operations, and a particular concentration on
the more responsive farmers through the Achikumbe‘ programs and irrigation settlement
schemes” (Nankumba 1981).

Four projects were established which followed the integrated intensive package
approach, the first of its kind to be financed by the World Bank (Lele 1978). Financing was
provided for construction of roads and social services as well as providing extension and

credit and marketing services. Emphasized cr0ps were cotton, groundnut and rice. However,

 

‘Achikumbe are progressive farmers who maintain the highest possible standards in their
farming without falling back for five years. They are awarded the prestigious Achikumbe
certificate.

.
5 These are described in section 2.3.2

prox

CO 111']

Sea0U. ad
Ma." 1937,
Private "ac
“19mm .i
”0:339 and

56mm“ Stu;

Mar
famers be

P .-
\u

 

17

the efforts proved to be to expensive and the yields did not increase as projected (Mkandawire
1985). There was little growth in the volume of marketed surplus. A new approach, the
National Rural Development Program (NRDP), was thus adopted where emphasis is placed on
provision of credit, extension and marketing services. Coverage was increased to the whole
country, and implementation has been incremental on a project-by-project basis.

Under the NRDP, the country is divided into 8 semi-autonomous Agricultural
Development Divisions (ADDs) (see Figure 2.1) which oversee the planning and delivery of
agricultural services to farmers. At the level of the farmer, services are organized as follows.
An extension service based on the training and visit system is available to all farmers. The
government provides a cadre of extension workers who teach and demonstrate crop and
animal husbandry and home economics topics. Each extension worker has approximately 800
farm families to reach. She divides her area into blocks. Farmers in each block meet with
her on a specified day in a rotation. Although these meetings are open to everybody, it is
mostly those who have received credit for inputs who attend. In principal, anyone can get
credit; however, because of requirement of credit worthiness, only those who have resources
for collateral receive credit. In all, only about 25% receive credit and patronize the block
meetings (Ngwira 1988).

Agricultural marketing is done by the Agricultural Marketing and Development Board
(ADMARC) through a chain of markets which purchase and sell crops. During the harvest
season, additional rural markets are opened to allow farmers to sell their crops easily. Since
May 1987, an act of parliament liberalizing agricultural marketing has made it possible for
private traders to buy and sell crops. This has created problems for farm households because
the traders have no incentive to go back to the rural areas and sell produce due to high
storage and transport costs (Kalua 1991). There are also problems for monitoring the food
security situation because it is difficult to follow the activities of private traders.

Marketing through ADMARC effectively imposes an export tax on smallholder

farmers because they receive only a small percentage of the world or urban market price even

18

 

"' s tors

 

 

 

 

 

 

.-.n....- —L-.

 

 

 

 

 

Figure 2.1: Map of Malawi Showing Agricultural Development Divisions and the Study Area

COI

CTO

T'UTE

scum a:
1“ IOtal

mOSlly L

19

considering administration and transportation costs. In contrast, estate producers sell their
crops straight to the international buyers and receive the world market price.

Most assessments of the Achikumbe and irrigation settlement schemes, integrated
rural development projects and the NRDP are negative (Lele 1975, Mphande 1984, Kydd and
Christiansen 1981, Mkandawire 1985). In the case of food crops, performance of the
smallholder sector with respect to maize production has been disappointing. The National
Sample Survey of Agriculture (1968 and 1981) show that yields have not changed. Until
recently, the country was regarded as an exception to the African food crisis, in that self-
sufficiency has been maintained in the main food crop, maize. However, the low productivity
and the increasing population have led to low per capita food availability. To date, low
productivity has meant that agricultural programs and policies have failed to ensure that the

majority of smallholder farmers are food secure.

2.3 Irrigation Potential in Malawi
2. 3. 1 Hydrological Resources

Malawi is a long, narrow landlocked country, some 840 kilometers from north to
south and varying in width from 80 to 160 kilometers. It is situated in Central Africa lying
between latitude 9 degrees and 17 degrees south and longitudes 33 degrees and 37 degrees.

Its total area is 118,484 square kilometers of which 24,208, or 20%, are covered by water,
mostly Lake Malawi. It is contiguous on the north and northeast with Tanzania, on east,
south and southwest with Mozambique and on the west with Zambia.

The country is aligned along the southern continuation of the east African rift valley
system, but it is mostly a plateau. The northern part of the rift valley contains Lake Malawi
which is 568 km long, varying in width from 18 km to 80 km. The River Shire, which'flows
from Lake Malawi in the south, takes a serpentine route proceeding through Lake Malombe to
the Zambezi River. Throughout its course, it provides many opportunities for gravity fed

water abstraction for irrigation (Malawi Government 1992). The Shire Valley from

Ka

val

the

\‘tpl
1.52
Mule
river:

theL

20

Kasinthula onwards is practically irrigable land because of easy access to water and the fertile
valley soils. This area is perhaps the single most under utilized resource in the country.

The plateaux are on the western side of the rift valley with the highest elevations in
the north and several mountain ranges in the central and southern regions. The Nyika and
Viphya Plateaux in the north, and the Dedza mountains and the Kirk range rise to between
1,524 and 2,440 meters in places. In the south, the Shire highlands, and the Zomba and
Mulanje mountain ranges rise to 2,100 meters. The plateaux are drained by many perennial
rivers which provide opportunities for irrigation in the flood plains and at their confluences to
the Lakes Malawi and Chirwa. Examples of this are Wovwe and the Likangala rice schemes.

The great variations in latitude and attitude are responsible for a wide range of
climatic, soil and vegetational conditions within the country’s comparatively small land area.
There are three climatic seasons. During the cool season, from May to August, there is very
little cloud cover. Temperatures in the plateau areas are 15.5 C to 18 C, and in the rift valley
20 C to 24.5 C. The coldest month is July, maximum temperature is 22.2 C and the
minimum 11 C. Light rain may occasionally fall on the higher southeastern and northeastern
slopes. The rain is sufficient to grow a second crop of beans or field peas. In September and
October before the rains start there is a short hot season with high humidity. Mean
temperatures range from 27 C to 30 C in the valley and 22 C to 24.5 C on the plateau.
During the hottest months, October to November, temperatures may exceed 37 C in the low
lying ares. The rainy season lasts from November to April, and over 90% of the annual
rainfall occurs during this period. Most of the areas receive an annual rainfall of 760 to
1,015 mm, but some areas on the plateau get over 1,525 mm.

The country of Malawi almost never experiences complete rain failure. Most of crop
failures are due to inadequate rain at critical times in the cropping season. But during the
1991/92 cropping season the country experienced the worst drought in more than 40 years
(Malawi Government 1993). The last famine of 1949, however, is considered worse in its

impact. The recent drought started in Southern Africa and spread to the whole country by

self-l
Maiat
Sparse
from 0

33‘: .cte

for

 

21

mid-January. The southern part of the country was the hardest hit. Rainfall was below half
of the normal level and food production dropped to less than half of the normal level.

In 1990/91, most parts of Malawi received less than half of expected rainfall in the
first half of the season (October to December) (Malawi Government, 1991). The 1992/93
cropping season started very well, but rains were interrupted up to mid-December. These

references to recent drought are used to illustrate the need for supplemental irrigation.

2. 3.2 Irrigation Policies

There are two kinds of irrigation establishments in Malawi: government supported
small scale irrigation projects and private projects. The latter can further be divided into two:
self-help small scale projects and large scale commercial irrigation. The government of
Malawi has not financed or organized large scale irrigation projects. The government-
sponsored projects are commonly referred to as settlement schemes. Tenants are brought in
from other parts of the country to work on land that has been developed by government
agencies.

In 1967, the Irrigation Branch was formed under the Ministry of Economic Affairs to
design, construct and maintain smallholder irrigation schemes. In 1968, the Settlement
Branch was formed to look after the settlement of farmers and to manage the irrigation
schemes. The aims were:

a) to prepare the country for the time that rainfed agriculture could not produce enough
food to feed the population; and
b) to make it possible to produce certain export crops in the wet and dry season

(Mphande 1984).

Figure 2.2 shows the schemes that have been developed by the Irrigation Branch and
the Chinese Agricultural Mission to Malawi. The schemes are located along the Lakeshore
plains and in the Shire Valley. The main crop grown is rice. The highest yields achieved by

farmers is 4.7 tons per hectare and average yields are estimated at 3 tons per hectare (the

    
   

ntzuzue

SCALE t- 1 : 4,000,000

 

Figure 2.2: Map of Malawi Showing Existing Projects Area

23

Table 2.2
Irrigation Development (Hectares)

 

Year Total Area Developed Area Planted
Developed within Year with Crops
1969/70 802 654 593
1970/71 1396 594 1288
1971/72 1841 445 1764
1972/73 2343 - 502 2094
1973/74 2590 2472 671
1974/75 2590 -- 2081
1975/76 3696 1106 2359
1976/77 3798 101 3078
1977/78 3899 101 3484
1978/79 3928 29 2750
1979/80 4016 88 2201
1980/81 4171 155 2546
1981/82 4171 --- 3852
1982/83 4171 --- 3548
1983/84 4171 -- 2377
1984/85 4535 364 3628
1985/86 4899 364 3819
1986/87 5273 374 4218
1987/88 5633 360 4506
1988/89 5903 614 600
1989/90 6054 421 4843
1990/91 6354 300 5083
1991/92 6654 100 5083
1992/93 7000 220 5349

 

Source: Malawi Government, Agricultural Statistics, Annual Bulletin (1993)

potential yield is 6 tons). The total land area developed is estimated at 7,000 hectares.
The Department of Irrigation conducted appraisal studies on the possibility of
upgrading self-help schemes and increasing the number of gravity-fed irrigation schemes
(Malawi Government, January 1992). Twenty-three and 18 sites were evaluated in Mzuzu
and Blantyre ADDs respectively. The evaluations were based on primary and secondary

indices.

24

The primary index included such criteria as presence of perennial source of water,
possibility of gravity irrigation, ease of abstraction, topography, drainage and soil fertility.
The secondary index captured experiences in committee organization, water management, and
communal work, and also economic considerations such as cropping intensity and increase in
production and costs of construction. The assessment shows that there is potential for up to
12 self-help projects to be redesigned and upgraded to assisted status in Mzuzu and Blantyre
ADDS.

Self-help irrigation projects are based on three tenets:

a) the projects should be farmer managed;

b) the area should not exceed 50 hectares;

c) abstraction of water should be by gravity (diversion): pumping or sprinkling is not
included.

Using the criteria above, it is estimated that another 2,000 hectares can be developed
on a self-help basis in these ADDs only. This excludes areas suitable for formal, larger
schemes as well as small areas where the parent river is too large to make small-scale
abstraction feasible. At present, 1,500 hectares of land are under self-help schemes in the
whole country, this is 7% of total irrigated land.

The irrigation projects that are envisaged in this study include cases where it is
financially feasible to pump water for irrigation, and are financed and run independently from
the government. Government assistance to self-help projects is a very sensitive affair, yet in
many cases this assistance is needed first to create the awareness of possibilities through
extension and demonstration. The second stage is to provide technical guidance, especially in
constructing intake works, but also in formalizing marketing channels for the produce. The
third and most sensitive step is financial support. These aspects are reviewed in the next
section.

Large scale commercial irrigation is used mostly in the estate sector for sugarcane

production. The major establishments owned by the Sugar Corporation of Malawi are at

25

Nchalo and Dwangwa. Some tobacco and coffee growing estate farms use sprinkler irrigation
to counter the effect of drought or to make possible early planting. The potential for
commercial overhead irrigation is technically unlimited since there are many perennial rivers
and suitable arable land. In many cases overhead irrigation could be used to grow off-season
crops on rainfed land. Only financial constraints can limit its use for production of some
crops. The total area developed by commercial agencies is about 16,000 hectares.

It is estimated that 290,000 hectares of land in Malawi can be irrigated by different
means, but only 7% of this has been irrigated (FAO 1986). This potential plus the valuable
experience the Ministry of Agriculture has had with irrigation makes investments in irrigation
a very effective way to address the problem of rain dependence and risky, low productivity
smallholder agriculture. However, some institutional issues may have to be addressed to

realize greater returns from investments.

2.4 Implementation of Irrigation Projects

There is considerable evidence that the potential gains from irrigation in developing
countries are far from being fully realized (Barnett 1984, Makhado 1984, Carruthers and
Small 1991). This evidence also shows that inappropriate institutions contribute to the
situation (Barnett 1984). This section provides a cursory review of these and other problems
that need to be addressed to improve chances of successfully implementing small scale
irrigation projects. The problems can be grouped into five.

The first relates to deficient design and technical management, which eventually leads
to insufficient delivery of water to farmers with negative impact on yields.

The second is inadequate financial resources, inappropriate institution arrangements
for financing projects, and low priority for maintenance of irrigation works. The mechanisms
for financing irrigation projects affects their performance (Carruthers and Small 1991). Issues
concerning maintenance are important for success of irrigation projects. If a project already

suffers from other difficulties, the use of technically insupportable technology guarantees

Cf
If!

a)

bl

Cl

 

 

26

failure. Morris (1984) cites a project in Mali where irrigation pumps were ordered from
India because they were considered technically optimal (the alternative was to get them from
France). But it proved difficult to order spare parts and find or train people to maintain them
in a timely manner.

The third, poor production techniques, has caused failure of many irrigation projects.
Irrigation projects which aim to reduce farmers’ exposure to naturally caused production risk
end up exposing them to institutional risks by participating in projects. For example, the

report for Mchenga Project pointed out the following problems:

a) farmers’ inexperience with the crops, in particular direct seeding, led to poor
germination;

b) lack of experience in pest'control led to 75 % loss of tomato crop in the first cropping
season;

c) farmers were wrongly sold by the National Seed Company choumollier instead of

rape, which resulted in undersupply of seed because of the difference in plant

population for the two (although almost the same) leafy vegetables. Choumollier has

lower plant population and the seed package contains less seed grains. The more
popular rape needs more seed per unit area. This led to a situation where some beds
meant for rape remained empty, and also to marketing problems since customers
prefer rape. The solution proffered farmers in this report, suing the company, is not
adequate.

There has also been a problem with cropping patterns. Crops were either under- or
over-watered. This problem was being solved at the time of conducting the study by
following independent blocks for every crop. The first method had a block for each farmer.
Since the sprinklers have a wide reach, it was difficult to control water for each crop on the
farmer’s block. Most of the problems at Mchenga and Diamphwe projects were exacerbated

by poor farmer supervision.

 

1183

irrlt

\

27

The fourth group of problems originate from low motivation of project staff. There is
need to reward irrigation extension officers based on criteria related to project performance.
This may be difficult if the extension workers are part of the Civil Service, hence the need for
irrigation projects to be autonomous from government.

Another disadvantage of public irrigation projects is that extension workers are
relocated every three to four years so that they do not feel responsible for the projects. It
also leads to discontinuity in rapport between farmers and staff.

The fifth problem concerns marketing. Mchanga and Diamphwe projects were
experiencing problems with marketing of leafy vegetables. Some farmers dropped out from
the project because of this. It is thus imperative to plan properly for marketing of
horticultural crops. In the first season, there was no market for lettuce. In the second
season, there was tough competition from dimba crops produced by other farmers. The
report of the Lilongwe ADD recommends:

intensification of market research to establish demand and supply as a guide on what
and how much to produce at a specific time period taking into account other
producers (Lilongwe ADD 1993, p4).

The sixth problem is poor water management. The result is usually inadequate water
provision to farmers. In rice growing projects, inadequate water management is held to be
the single most important factor in explaining the gap between actual and potential yield (FAO
1986). A simple calculation from the figures from Mchenga and Diamphwe projects show
that only the best farmers (with highest yields) should participate in the project if it is to be
independently financially viable. There is need to reduce constraints to increasing yields.

Finally, irrigation projects experience problems if they are based on assumptions
reflecting inadequate understanding of social relations of production. This is particularly true
in the area of labor supply. Project officers tend to deal with men or use abstract concepts of
the household. "Planning documents frequently contain calculations which involve counting
of person-days required to produce a certain crop mix and reconciling this with the available

person days. This is a crude and destructive method which deforms reality" Barnett (1984).

Women

can leat

11183 4:1,-

mitts

28

Women tend to be ignored by planners or taken for granted. In some cases, targeting women
can lead to more efficiency:

if irrigated rice plots and the whole technology of growing rice together with credits
originally given to men by the Taiwanese and World Bank programmers had been
made to women as well as men, it is probable that double cropping of irrigated rice
would have been achieved in women’s fields at least. Women are in a much stronger
position than men to cultivate irrigated rice in the rain season as labor demands fit in
with customary sexual division of labor (Dey 1980).

2.5 Summary

This chapter reviewed agricultural and irrigation policies in Malawi, to provide a
macro-context for evaluating the objectives, methods and findings of the study. The chapter
emphasized the negative synergistic impact on household food security of poor performance of
smallholder agriculture; increasing population pressure; and a small manufacturing sector. It
is argued that the risky nature of rainfed, low input cultivation makes irrigation an attractive
prospect. The country has suitable hydrological and land resources, as well as valuable past
experience to enable implementation of financially feasible irrigation projects. However some
institutional reforms and innovative management may be necessary to ensure success of

projects.

m
kt
dn
dd
part

farm

COUm
Emu
imPOn
Smdh.
Menu
laclot

Hdhm,

CHAPTER 3: ANALYTICAL METHOD

This chapter discusses the rationale for the selection of analytical method used in this
study. It begins with a brief review of smallholder farmer objectives and decision making in
risky environments, to justify the selection of a linear programming model based on Target

MOTAD with Telser’s safety first probablistic criterion.

3.1 Smallholder Farmer Objectives

Ellis (1988) provides a lucid review of the several theories posited in agricultural
production literature about the economic behavior and goals of smallholder farmers. The
following theories are included: the profit maximizing peasant; the risk averse peasant; the
drudgery averse peasant; the farm household peasant; and the share cropping peasant. His
definition of peasant is also our concept of ’smallholder farmer’ as "family farmers only
partially integrated into incomplete or imperfect markets. The threefold emphasis is on
family, on partial engagement in markets, and on the imperfections of those markets ", p181.

Schultz (1964) was among the first to suggest that smallholder farmers in developing
countries might be optimizing economic agents. His hypothesis of smallholder farmers being
’efficient but poor’ has been researched and evaluated extensively. This hypothesis derives its
importance not from accurately describing farmer behavior, but from its success in placing
smallholder farmer rationality fim'ily on the research and policy agenda. Prior to Schultz the
literature on traditional agriculture was permeated with stereotypes of laziness, perversity,
lack of motivation and irrationality on part of peasants as economic agents.

The Schultzian hypothesis ascribes to peasants the motive of profit maximization.

Profit maximization has both a behavioral content and a technical-economic content. Most of

29

30
the research in this area has rejected the efficiency hypothesis in its pure form in favor of
some form of constrained profit maximization (Lipton 1968, Ray 1985).

In most economic modelling, utility maximization is equated with profit
maximization. This is correct if farmers have no other objectives and are operating in risk-
free environments. Empirical work shows that economic agents’ desires to maximize profits
are often constrained by the need to avoid or manage risk. Research studies support the
proposition that small farmers are risk averse (Wolgin 1975; Schluter and Mount 1976; Dillon
and Scandizzo 1978; Binswanger and Sillers 1983). Maximizing utility under uncertainty may
result in lower use of resources, lower levels of returns and diversification of activities
(Samuelson 1967, Sandmo 1976, Brink and McCarl 1978, Weirnar 1988).

An important problem in the empirical literature is attributing to risk aversion all
departures from economic efficiency. This is especially true in studies that use variations to
objective data, like prices and rainfall patterns, as the basis for drawing conclusions on the
subjective behavior of farmers (Binswanger 1980, Binswanger and Sillers, 1983). So, for
example, mixed cropping has been explained as a response to risk when actually it may also
' be a way of maximizing returns to labor and land.

In this study, smallholder farmers are perceived to be utility maxirnizers who are risk
averse. The most important type of risk is yield risk (and hence income risk) as affected by
rains and augmented by market imperfections. Risk aversion plus cultural preferences and
taste considerations for local varieties of maize lead to farmers trying to produce their local
maize consumption requirements (Peters and Herrera 1989, Smale 1991 and Kydd 1989).
Thus farmers exhibit a safety first response to uncertainty. However, farmers may fail to
optimize in many cases perhaps because of cognitive (relating to the way problems are
structured and the strategies), not necessarily motivational (relating to the amount of

emotional effort expended) (Schoemaker 1982), limitations. This is one reason for well

31

informed extension advice based on results of relevant smallholder agricultural production

studies.

3.2 Decision Making Under Uncertainty

3.2.1. Risk and Uncertainty

Risk refers to situations where probabilities can be attached to the occurrence of
events. Uncertainty refers to situations where outcomes of activities are stochastic but there is
lack of information on probabilities of occurrence of outcomes (Knight 1921). This definition
ushers in the philosophical problem of what probabilities are. Probability lacks primary
sensory evidence as to its existence and may be an invention rather than a discovery.
Churchman (1961) expressed the problem this way, "almost every one knows what it means
to say that the event is probable-except those who have devoted their lives to thinking about
the matter. " Additionally, there is evidence that people from one culture may perceive
uncertainty more in probablistic terms than those from another culture (Hacking 1975,
Lawrence and Wright 1977).

Uncertainty characterizes the environment confronting farmers. This environment
contains a wide variety of uncertain events to which farmers will attach their subjective
probability beliefs. Economic analyses are sometimes based on the decision maker’s personal
belief about the probability of occurrence of uncertain events rather than objective
probabilities. In the case of rainfall, for example, what is important is not the known past
average occurrence of drought, but rather the farmers’ personal view about the likelihood of
drought.

This distinction between objective and subjective probabilities is important for two

reasons. First, historical data may be an unreliable indicator of future events. Second, in

32
many cases if a disaster strikes, people react as if it will strike again sooner than evidenced by
historical data. Their behavior is not based on historical data but on most recent experiences.
This is consistent with the concept of reinforcement in behavioral psychology (Schoemaker
1982). It is this manifest decision maker behavior that matters in risk modelling because it
affects farmers’ decisions about resource allocation (Ellis 1988). The distinction between
objective and subjective probabilities is also important for evaluation of the expected utility

hypothesis that undergirds the decision making models reviewed and used in this study.

3. 2. 2. Models of Risk Analysis in Agriculture

The decision making process under uncertainty can be reduced to five steps (Crawford
1992):

1) Identification of the set of alternative actions. Actions X,,...,Xj should be mutually
exclusive and exhaustive of alternatives available.

2) Identification of states of nature which may occur.

3) Assignment of probabilities to the states of nature. These are degrees of belief held
by the decision-maker of the likelihood of each state of nature occurring.

4) Specification of the outcomes of each activity under the various states of nature.

5) Choice of criterion for deciding the optimal plan.

When there many possible farm plans to choose from, a farm manager can be guided
by what proved successful in the past for herself and others. She will adjust her program
marginally using some informal arithmetic or a more or less systematic budget approach.
However, there are more structured methods to solving risky decision problems. They range
from using conservative estimates for certain random variables (e. g. price and yield), through

methods which explicitly incorporate density functions for the random variables. With

Wha-

bl

mar 21f]

fi

acIll-ill:

33

increased computational and software facilities these models can be easily solved by
researchers and decision makers through mathematical programming.

Mathematical programming allows the testing of a wide range of alternative actions
and analyzes their consequences thoroughly with a comparatively small input of time.
Questions relating to proposed changes to a farm plan can be answered rigorously and quickly
(Beneke and Winterboer 1973). A general linear programming model to represent optimal
farm decision making process is:

max 2 = l'Jcj X].
such that:

E 3,-ij < b,, for all i e [1, m]

xj > o, for allje [1, n]

where:

Z = total gross margin

Xj = the level of the jth farm activity, and n is the possible number of activities
c,- = the expected gross margin per unit of activity j

a,- = the amount of resource i needed to produce a unit of j

bi = amount of the ith resource available

The problem is to identify the mix of farm activities that maximizes the total gross
margin Z, using no more than the resources available, and not involving negative levels of

activities. In matrix form the problem can be presented in this way:

2'”

T “\J

S

 

34
Table 3.1

A Simple Linear Programming Table

 

 

 

 

Row name

X1 X2 ..... X11 RHSa
Objective function Cl Q ..... Cn MaxiFu'ze
Resource constraints
1 all al2 ..... aIn < bl
2 a21 a22 ..... a2n < b2
m aml am2 ..... amn < bm

 

a/ RHS = right hand side

In ordinary analysis, the parameters C], aij and b, are assumed to be known
with certainty. However in agricultural production, activities have uncertain outcomes owing
to the biological nature of enterprises, variable weather and environmental conditions,
changing demand and unpredictable government policies which affect prices. All these lead to
yearly or seasonal variation in incomes. Under these circumstances, the ordinary LP does not
give optimal results because the decision criterion used in ordinary LP does not take into
account the riskiness of alternative plans.

Many criteria are available for risk analysis problems. The choice of decision
criterion for a particular risk analysis problem depends primarily on the following: a) the
nature of risk being dealt with, e. g. objective function or constraint set risk; and how the risk
is defined for programming purposes, e. g. deviations below target or variance around mean;

and b) how the farmers are perceived to respond to risk e.g. safety first response.

3.2.1

found
mode}

lmpro

an ord:
hyporh

attort<

 

ordert:
prefers I
prefer:
if X; is
x) an:
that a rl

“that

albttrart

 

“mm \)

funnier

ACcord

 

35

3. 2. 1. 1 Direct Expected Utility Maximization

Most of the risk analysis criteria that use mathematical programming have their
foundations in the expected utility model (EU). The first group uses the expected utility
model directly and the second group, cognizant of the weaknesses of the model, attempts to
improve on it.

The expected utility hypothesis postulates that under certain assumptions or axioms,
an ordinal utility function exists by which prospects can be ranked. The foundation of this
hypothesis as developed by von Neumann and Morgenstem (1953) consists in a set of four
axioms which are: ordering, transitivity, continuity and independence of prospects. The
ordering axiom requires that for any of the prospects X, and X2 the decision maker either
prefers X, to X2, X2 to X, or is indifferent between them. Transitivity implies that if X, is
preferred to X,, and X2 is preferred to X,, then X, is preferred to X3. Continuity implies that
if X, is preferred to X,, and X, to X,, then there is a mixture of X, and X3 that is preferred to
X,, and a mixture of X, and X3 over which X2 is preferred. The independence axiom says
that a risky prospect X, is preferred to a risky prospect X2 if and only if a p chance of X,
with a (l-p) chance of X3 is preferred to a p chance of X2 with a (l—p) chance of X3, for
arbitrary positive probability p and risky prospects X,, X,, and X3. It is the independence
axiom which gives the EU model theory its empirical content by placing a restriction on the

functional form of the preferences by implying that it must be linear in probabilities.

According to the EU model, for any utility function U(X):
a) if a risky action, X, is preferred to another X,,
then U(X,) > U(Xz), and
b) U(X,) = EiU(Yij) = EpiU(Yi,-).
The optimal X,- is the one that maximizes expected utility.
EU(X,-*) = Max [U(X,)] = Max [Emma].

CK

.«e
b\

36

where U(X,),...,U(X,) = utility of choices X,,...., X,-

X,‘ = ex ante optimal choice among actions;
p, = probability of state of nature S, occurring; and
Y,- = the outcome of Xj when Si occurs

Alternative actions are thus ranked according to the probability of the states of nature and the
relative preferences concerning the outcomes as embodied in the utility function.

The foregoing definitions enable the illustration of key concepts in risk analysis
(Figure 3). First, risk aversion leads to a concave utility curve so that the utility of the
expected monetary value of a risky plan is higher than the expectation of the utility of the
action. Risk averse decision-makers maximize expected utility, not utility of the expected
value of the risky plan. The certainty equivalence (CE) is the amount that will give the same
utility as the risky plan—EU(Y). The risk premium is the difference between the monetary
outcome of a risky plan and the CE. This amount will make the decision maker indifferent
between receiving the certain amount, CE, and taking the risky plan. The risk premium is
positive for risk averse individuals, and they prefer certain outcomes above the CE to the
risky action, and prefer the risky action if the certain outcome is below the CE.

The degree of risk aversity can be measured by the Arrow-Pratt absolute risk aversion
function

17 = -U"(Y)/U’(Y)

The direct application of expected utility maximization requires assumptions
concerning the nature of the utility function. Tobin (1958) used the utility function

U(y) = (1+by) +by2
for which maximizing expected utility is equivalent to maximizing:

(1+b) u + b(02+u2)

Freon“,

utility is e,

 

Where [a

37

 

 

“Em

sum

 

 

 

 

um)

 

 

 

 

 

 

r1 D! Em '2 V

Elma: llhslutloumemunlnlsumslon,
Certainty Esau-lean. and an Premium

Freund (1956) proposed the use of negative exponential utility function
U(y) = 1- e "V
where a > 0 a is measure of the decision-maker's attitude toward risk. In this case maximizing
utility is equivalent to maximizing:
u - a02/2
where u = mean of y

o = standard deviation of y

rat:
ext

cles

$011

and

38

Another approach to direct expected utility maximization is the Exponential Utility
Moment Generating Function (Lambert and McCarl 1985; Collender and Chalfant 1986;
Babcock, Chalfant, and Collender 1987). It is based on the assumption of an exponential
utility function, and the maximization of expected utility is based on the moment generating
function for the random variable. In some variations of the model the probability distribution
of the random variable has to be specified. Each type of distribution yields a different
moment generating function.

Although the expected utility hypothesis is consistent with economists’ concept of
rationality, empirically it holds as an exception rather than rule. Schoemaker (1982) has
extensively reviewed research that evaluated the model. He concluded that the model fails to
describe how people behave: They do not structure problems holistically and
comprehensively as the hypothesis suggests; they do not process information, especially
probabilities according to the EU rule; and it poorly predicts results for experiments. Another
difficulty is related to the existence and nature of probabilities, and whether they are
subjective or objective.

Research results tend to agree that many careful, intelligent decision makers do violate
some axioms of expected utility theory even upon reflection of their choices (MacCrimmon
and Larson 1979). The independence axiom has been shown not to stand up to the data.
There are four systematic violations of it: the common consequence effect, the common ratio
effect, over sensitivity to small probability changes and the utility evaluation effect (Machina
1985).

So the EU model holds only in exceptional cases, with well trained decision makers,
for well structured tasks. It is thus worth while to explore the option of considering
modification to standard theory. Loomes and Sugden (1981) and Fishbum (1981) have
developed alternatives that drop some of the axioms, and Machina (1982) has shown that

major concepts and tools of EU do not depend on the independence axiom. Part of the

proble:

langua,

(fried:
corntra
urns
The ris'.
provide
preferer

0116105

3.21. a

damn,
Criteria:
distribu,z
bi the El
demOmt:
”1333ng
”lemma
“Gm \
quadl’allc
Tl‘.
Sllch max

I"
4..

 

39

problem in expected utility hypothesis stems from lack of consensus on how to encode
language and context concerning preferences, for investigation.

Although some results of empirical work are still consistent with the hypothesis
(Friedman and Savage 1948, Tobin 1958), its use in empirical applications continues to be
constrained by the need to define suitable utility functions. The knowledge of this is not
always available. This has led to a search for other ways to deal with risk in programming.
The risk efficiency approach is the most common one taken. Risk efficiency criteria can
provide a partial ordering of alternative courses of action for the decision makers whose
preferences conform to a specified set of conditions placed on the utility function. The first

one to surface was the Mean-Variance criterion.

3.2.1.2 Mean-Variance Analysis

The mean-variance (E-V) criterion is based on the proposition that for any two
distributions of equal means, a risk avoider will prefer the one with smallest variance. The
criterion is not always consistent with the EU model. When decision variables are normally
distributed or the distributions differ only by location and scale, the optimal results selected
by the E-V and EU will be the same (Meyer 1987). Levy and Markowitz (1979)
demonstrated that solutions to the expected utility maximization problems are equivalent to
maximization of a linear combination of the mean and variance. This corresponds to the
maximization of expected utility of income when either the utility function is quadratic or the
random variable is normally distributed. Thus the application of E—V criteria results in a
quadratic programming model of the form

max Y - d> 0v2 = 2c, X,- - ¢220,,X,X,
such that:

2a,, xj s b, (i= 1, ...,m)

xj 2 0 for all j

rtplzc;
ddttr.
$th 1

Tl'lCLl.

ograr
US tr.
tse b,
90).
l5 at

lies

h re:

‘1‘.
Kit“;

1.3

 

l at:

 

40

The objective function maximizes expected total gross margins less a risk aversion parameter
(p) times the variance of total gross margins. If the gross margins are normally distributed, a
= 2d) the coefficient of absolute risk aversion parameter in the exponential function U(y) =
1- e "3' (Freund 1956). Another formulation is to minimize variance subject to a given level
of income (Markowitz 1952). The difference between the two is that solving the first one
requires specifying ¢ while in the second, mean income has to be specified.

E-V analysis leads to quadratic utility functions which have the theoretical
implications of increasing absolute risk aversion to which there are objections (Arrow, 1964).
Additionally, empirical work shows that crop yields have negatively skewed distributions such
as the gamma or beta (Day 1965) making it difficult to uphold assumption of normality in
agricultural production modelling. This has significant implications for the study at hand
because yield variability is the major cause of uncertainty. When implementing E-V
programming the mean and variance of the population has to be estimated from a sample.
This may introduce bias leading to selection of optimal decisions that differ substantially from
those based on true population, if distributions of variables are not normal (Chalfant et al
1 990). Levy and Markowitz (1979) have also argued that this criterion is acceptable where

risks are small relative to total wealth. Being based on the EU model, the EV criterion

carries over some of the same weaknesses.
The use of variance as a measure of risk in E-V analysis implies that both low and

high returns are undesirable. Yet empirical literature shows that decision makers frequently
associate risk with failure to attain a target return (Patrick et a1 1985, Peters and Herrera

1 989) . The mean-semivariance criterion was developed to redress this problem.

3. 2- 1.3 Mean-Semivariance (Lower Partial Moments)

Porter (1974) has shown that efficient sets selected using semivariance around the

mean are more closely aligned with second degree stochastically dominant efficient sets, than
«3 E—V efficient sets. He also showed that when risk efficient sets are calculated about a

 

fixed j

fUIICIIt

where

utility

3.2.1.

It is us“
allcmat
dominai
Pmpct‘ti
mm]

due to I»

 

 

41

fixed point (versus mean) they are consistent with maximizing utility, where the utility
function, U(Y), of a random variable Y, is of the form:
U(Y) = a + bY + c{(Y-h)2} ifY < h
= a + bY if Y > h.
where a, b, and c are constants and h is the fixed reference point of Y. This semi-quadratic

utility function is consistent with the premise that high returns are not inherently undesirable.

3.2.1.4 Stochastic Dominance

Stochastic dominance is a major risk analysis tool in agricultural production studies.
It is used independently to select optimal plans, but also to evaluate the robustness of
alternative risk analysis criteria. Stochastic dominance permits the elimination of clearly
dominated distributions for all decision makers whose utility preferences exhibit certain
properties. It uses the whole income distribution, not just the mean and variance. There are
several types of stochastic dominance analyses: first and second degree stochastic dominance
due to Hadar and Russell (1969) and Hanoch and Levy (1969); third degree stochastic
dominance (Whitrnore 1970), stochastic dominance with respect to a function Meyer (1977),
and convex set stochastic dominance (Fishburn 1974). Only the first two will be discussed.

The first degree stochastic (FSD) ordering rule for two risky cumulative probability
distribution functions F(Y) and G(Y) is: F dominates G if and only if F(Y) s G(Y) for all y
with a strict inequality at least in one value. FSD imposes the restriction that the utility
function must be monotonically increasing, and its first derivative, U’(Y) should be greater
than zero. Thus the marginal utility of the maxirnand should be positive. This restriction
defines a class of decision makers who prefer more to less income. It can accommodate
decision makers characterized by Arrow-Pratt risk aversion parameters varying from negative
infinity to positive infinity. However, FSD is limited in its ability to order action choices
since most choice sets have few alternatives whose cumulative probabilities functions do not

cross .

on the
maxim
decisio

PfOtlllCt

the mo
distrib:
ignore:
(Anders
cornea:

the effz.

3.2.1.5

Criterttr

Hillel.

ll miter:

 

 

42

For Second Degree Stochastic Dominance (SSD), an additional restriction is imposed
on the utility function that, the second derivative, U”(Y) < 0. The marginal utility of the
maximand should be increasing at a decreasing rate. It identifies the class of all risk averse
decision makers. The risk aversion parameter can vary from 0 to positive infinity. SSD
produces smaller efficient sets than FSD.

E-V efficiency is neither a necessary nor sufficient condition for SSD efficiency but
the two are equivalent when the utility function is quadratic (Tobin 1958) or all probability
distributions are normal allowing the moments of distribution higher than the second to be
ignored (Samuelson 1967). Stochastic dominance is consistent with the EU hypothesis
(Anderson 1974). Its major limitation is selecting efficient alternatives through a pairwise
comparison process, and that FSD and SSD tend not to eliminate many action choices from

the efficient set.

3.2.1.5 Minimization of Total Absolute Deviations (MOTAD)

The MOTAD, developed by Hazell (1971), is mean-absolute deviations (E-A)
criterion which has an important advantage over the EV criterion in that it leads to a linear
model. The following utility function is assumed in MOTAD:

U(Y) = a + Bx- clY- E(Y)|
It minimizes total absolute deviations subject to the usual constraint set and, in addition,
subject to a specified mean income level.

It is represented as follows:

Min Vy‘

Such that:

AY s b

RY = E

(R-R‘)Y + Iy‘ - Iy+ = 0

Y, y‘, y”, 2 0

Chit
Smal

altra

lob,

[mes
berte:

erfor

43

Where:

V = l-by-s vector in which each element is 1 and where s is the number of years
(observation of states of nature) considered

A = m by n matrix of technical co-efficients, where m is the number of constraints and

n is the number of activities

Y = n by 1 vector of activities

b = m by 1 vector of resource levels

y+ = s by 1 vector of annual positive income deviations from Y
y' = s by 1 vector of annual negative income deviations from Y
E = mean income

r = a 1 by n vector of expected income for each activity

R = an s by 11 matrix of income for each activity

R' = an s by 11 matrix consisting of s r vectors

I = an s by n identity matrix

0 = a column vector of appropriate length composed of zeros.

Hazell, who developed the MOTAD as an alternative to E-V and mean-semivariance
criteria, pointed to the advantages of computational ease, the model being linear and requiring
smaller matrices. Defining risk as negative deviations also makes the MOTAD criterion more
attractive than the E—V criterion since only negative deviations need be minimized. If the EA
is an acceptable representation of farmers utility function, MOTAD offers a single alternative
to both the quadratic programming of the EV criterion and the semivariance models for all
types of gross margin distributions. Also it leads to a linear model of farm plans, making it
better adapted for post-optimality analysis.

Objections to the MOTAD include the charge that it introduces a lot of estimation
error in the variance. But Johnson and Boehlje (1981) have shown that if Y, and Y, are
symmetrically distributed with the same means: and density functions F,(Y) and F,(Y) are
such that F,(Y) 2 F,(Y) everywhere in the region u—d< Y < u+d with at least a strict
inequality in one interval in the region, and else where F,(Y) > F,(Y) then F,(Y) will have
smaller variances and smaller MOTAD variance estimates. Thus the MOTAD may introduce

varia

MOT

lete
it is
theo
line

porn

3.2

10 n
taco

SCH}

llbet
The
Max:

SUE}:

44

variance error but the ranking of prospects by quadratic programming will be the same as by
MOTAD.

The MOTAD model does not have direct linkages to the EU. But since it was
developed as an alternative to the E-V criterion which has EU model theoretical foundations,
it is thought to carry over the weaknesses of EU model (Boisvert 1988, 1990). Recent
theoretical developments indicate that MOTAD can be improved to bring its results more in
line with other acceptable decision criteria. For example, comparing or calculating risk from
points (means) that are not equal is misleading. Watts et a1 (1984) have shown how MOTAD
can fail to discriminate between two choices where one is clearly dominated using SSD or the
Target MOTAD. Hazel] himself acknowledged that the MOTAD "can not be rigorously

justified as a substitute of quadratic programming in deriving E-V plans" (Hazell 1971).

3. 2.1. 6 Target Minimization of Total Absolute Deviations

Target MOTAD due to Tauer (1983) is a negative absolute deviation model. It seeks
to maximize expected returns while making sure the sum of negative deviations from a target
income level is kept below a specified minimum. It contains some elements of the mean-
semivariance and MOTAD. The following utility function is assumed:

U(Y)=a+bY+cl(Y-h)| ifY<h

= a + bY if Y > h
where U(Y),Y, a, b, c and h are defined as before.

The model is represented as follows:

Maximize RY

subjectto:
AY<b
R+Iy’+q>0
vy’-d=0

Y,y’>0

45

 

Where:

q = s by 1 vector in which all elements equal a fixed target of the random variable of
interest; 5

d = the maximum acceptable total negative income deviations from the target;

y- = is an s by 1 vector of annual negative deviations from the target and all others

defined as above.

The target MOTAD frontier is developed by parameterizing d.

Minimizing negative deviation from a target is more congruent with 1) actual

behavior of decision makers, 2) Von Neumann-Morgenstem utility functions, and 3)

 

stochastic dominance relationships [Fishburn (1977), Holthausen (1981) and Porter (1974)].
Target MOTAD improves on the MOTAD by increasing the discriminatory power
through using a target rather than the mean as a reference point for risk. It also uses a
criterion that is more consistent with reality, i.e negative returns are the measure of risk. The
E-V and MOTAD imply that the only way to reduce risks is to reduce income. Target
MOTAD does better on this. One disadvantage of Target MOTAD is that the both of the
attributes being parameterized are in the constraint set and so may require more time and

resources .

3. 2.1. 7 The Safety First Approach

Safety first is defined as behavior in which the probability of failing to achieve farmer
goals impacts and constrains the activities undertaken (Atwood 1988). It is more a way of
defining risk than a method of risk analysis. However once risk is defined in safety first
concepts, it has operational implications for the specification of risk models used. Various
specifications of the safety first behavior are available but all are concerned with the
probability of failing to achieve farmer goals. The probability of concern is expressed as:

Pr( Y < g ) < p
where Y is a random variable like income, g is a given goal of the firm, and Pr(.) is the

probability of the prospect, and p is upper limit on probability.

 

46

Examples of the safety first criterion are:

a) Min Pr (Y < g) [Roy (1952)]

b) Max E(Y) subject to Pr ( Y < g ) < p [Telser( 1955)]

c) Max Y subject to Pr ( Y < g ) < p [Kataoka (1963)]

where E(Y) is expected income, Y is income and g is the goal, usually the lower limit or
disaster level of income.

Chance constraints were initially used to deal with random coefficients in the
constraint set (Chames and Cooper 1959). Imposing safety first chance constraints in
optimization models can be difficult if the income distribution is non-normal. An alternative
is to translate the probablistic constraint into a deterministic equivalent constraint. This is
difficult if the random variable is endogenously determined in the model (Sengupta 1969).
Another approach utilizes the assumption of a multivariate normal distribution and the ability
to generate a E-V set of solutions (Pyle and Tumovsky (1970). Musser (1981) employed the
assumption of normal distribution of the stochastic variable to use the expected gain-lower
confidence limit criteria due to Baumol (1963) to implement a safety first model. These
methods require knowledge of the cumulative or probability density functions of the stochastic
variable. This knowledge is frequently not available.

The use of a stochastic inequality to impose safety first probablistic constraints has
been considered by several authors, e. g. Anderson, Dillon and Hardaker (1977); Sengupta
(1969); Berck and Hihn(1982). The first two used a Chebychev-type inequality to calculate
the upper limits on probability of failure. The inequality is:

Pr(IY-p|> k0) < 1/1c2
where Y is a random variableu is the mean, k is some constant and a is the standard
deviation. It states that with any probability density function the probability of a random
variable Y falling more than k standard errors from the mean u, is less than the inverse of
k2. If the distribution is symmetric, a modified version is used

Pr(Y < ,i - ka) < 1/2k2

 

47

The upper limits so derived using Chebychev-type inequalities are generally
overestimates which lead to selecting a more conservative activity mix than is necessary when
imposed in optimizing models. Also, being nonlinear, they can be quite difficult to
implement. Berck and Hihn (1982) introduced an inequality which offers the potential to
provide less conservative probability limits than the Chebychev inequality if the population is
asymmetrical.
This inequality is:

Pr(Y < u-—m¢r)<1/m2

where m is some specified reference point.

3.3 Telser’s Safety First Probablistic Criterion in Target MOTAD

Atwood (1985) and Atwood et a1 (1985) adapted this inequality to improve safety first
probability limits. The following inequality was used:
Pr[Z < t- LQ(a,t)] < 1/L“

Q(a.t) = E(t-Z)“ fir

fi = probability density of Z, the random variable, in state i

t = reference level or target below which deviations are measured
a = some constant

L = probability limit

If a = 1, Q(a,t) becomes a linear lower partial moment (LPM) and the equation above
becomes:
Pr[Y < t- LQ(t)] < 1/L
Atwood later illustrated how enforcing a sufficiency condition in a Target MOTAD model
guarantees that the inequality is satisfied (Atwood et al 1988). The sufficiency condition is:
t— L‘Q(t) > g
where:

t = target income endogenously selected, and

 

48

g = goal of the firm or farm
Q(t) = is sum of deviations measured from t; and
L* = l/L, the reciprocal of the probability limit

This particular version of the sufficiency condition corresponds to Telser’s safety first

criterion. The Target MOTAD with Telser’s safety first criterion can be represented as

follows:

max RY

subject to:
AY < b 3.3(a)
R+Iy'+q20 3.3(b)
vy‘ - Q(t) = 0 3.3(c)
t- L'Q(t) 2 g 3.3(d)
Y, y' 2 0

(All parameters and variables defined as before.)

This method will select the activity mix which optimizes the choice variable while
simultaneously enforcing probabilistic constraints upon the deviations of income from the
optimal expected income with the least constraining LPM stochastic inequality.

The method differs from the plain Target MOTAD in two main ways. First, the
target, t, is no longer the same as the goal, but it is set endogenously by the algorithm at a
level which satisfies the sufficiency condition, t - L*Q(t) 2 g. Simultaneously, 3.3(b) and
3.3(c) compute the corresponding weighted deviations. If the sufficiency condition is
constraining, the level of t selected will correspond to the least constraining LPM possible.
This is the same as selecting the least constraining LPM from the set of LPMs for which t -
L*Q(t) 2 g.

Atwood’s LPM method is superior to the plain Target MOTAD with respect to the
level of difficulty in making a prior determination of some parameters. While the plain
Target MOTAD requires selecting a target for the choice variable and setting an upper limit

on deviations, with Atwood’s method, the modeler must choose the goal and level of

49
probability limit. Making these choices reduces the set of target incomes and expected
deviation levels to be considered because the algorithm selects the target (McCamley and
Kliebenstein 1985). Also, the goal and probability limit can be set more objectively than a
target income or minimum deviations.

Imposing a safety first constraint does not reduce the number of questions to be
answered, but the questions may be easier to answer. In addition, the method will not always
find the solution that maximizes expected income subject to the upper limit on probability. It
may fail because the safety first constraint involves a trade-off between a target level and
expected deviations from that target. Another weakness of the model is that if the decision
maker is concerned only about the probability of below goal incomes, and not the magnitude

of the deviations, then a different approach would be appropriate.

3.4 Conclusion

Target MOTAD with Telser’s safety first probabilistic criterion using Atwood’s LPM
is selected for this study because it mixes the desirable elements of several decision criteria,
viz: the safety first concept of risk and measuring risk as negative deviations from a target.
Also, Atwood’s LPM does not require the knowledge of the mean or variance of
distributions. This is advantageous in this study because the assumption of normality may be
difficult to uphold. Because it is a linear model, it is easier to implement, and its dual

information has full decision making interpretation unlike those of quadratic models.

CHAPTER 4: DATA COLLECTION

Data collection activities for this study began at the end of September 1992. A survey
was conducted between November 1992 and February 1994. The data sought included labor
available to households and allocation to crop production activities, use of land and size of
gardens. Several government offices were contacted to prepare for the survey and get useful
secondary data. The Department of Irrigation (DI) and the Blantyre Agricultural

Development Division (BLADD) management were most important in this regard.

4.1 Selection of Study Area
The study was conducted in Chiradzulu North Extension Planning Area (EPA) of

BLADD. This is an area in the Southern Region of Malawi, part of the Shire Highlands.

(See Figure 2.1). This EPA was selected for several reasons:

a) It has a high percentage of farm households with small land holdings compared to
other regions of the country, and these are the households most in need of intensive
agriculture. ,

b) It has topographical and hydrological resources that will permit irrigation.

c) Prior food security studies have been conducted in the EPA and an adjacent EPA,
Zomba South (Center for Social Research 1990 and Peters and Herrera 1989, 1992)
which makes it easier to design the study and validate results.

(I) The researcher is familiar with the area.

4.2 Sampling

We had initially decided to interview those households with dimba gardens in the EPA
who had been sampled in the Market Liberalization Study carried out by the Center for Social
Research, University of Malawi. Their sample was a random one and as a result there were

very few households with dimba gardens. To get enough households with dimba gardens in

50

 

51

the EPA we resorted to purposive sampling. This was done with the help of the field
assistants in the EPA, each of whom is responsible for a section. The field assistants were
asked to furnish lists of farmers who had dimba land. However, the numbers were not
sufficient to perform stratified random sampling, so it was decided to interview all those
farmers with dimba land in order to meet the sample requirements. The households were
divided amongst the enumerators who interviewed enough farmers to fill the quota for a
section. Where present the heads of households were interviewed. In a few cases
arrangements had to be made for a visit.

A total of 270 household heads were interviewed, 19.7% of which were female.
Several studies show gender to be strongly related to food security variables (Msukwa 1990,
Chipande 1987, Peters and Herrera 1889). The proportion of female headed households in
this particular area is 30% to 40% (Malawi Government 1981). Although the aim was to
stratify the survey sample accordingly, it appears that the share of female headed households

among farmers with dimba land is considerably less than 30%.

4.3 The Questionnaires

A structured questionnaire was used to collect data consistent with the objectives of
the study. It had three parts: the household questionnaire; the enterprise questionnaire; and
the garden measurement sheet (see Appendix B). In all cases the respondent was the head of
household or spouse. The household questionnaire contained sections on labor availability,
patterns of land use, use of credit and extension credit services and food security strategies.
It also included open ended questions on willingness to participate in and pay for irrigation.

The enterprise questionnaire was administered for every plot‘5 of land on which there
was a different crop mixture. Thus, several enterprise questionnaires could be administered

for a parcel of land. The information collected through this questionnaire included: labor use

 

6 In this study a plot is defined as a piece of land with same the crop mixture. A parcel is
a contiguous piece of land which may have several plots. In some cases a plot and parcel may
be the same.

52
for the various activities on the plot, quantity of inputs used, (such as pesticides and
fertilizer), crops grown, quantity harvested and sold, and sale prices.
The garden measurement sheet was used to calculate area of the plots. Since most of
the gardens are polygons, the length and angles for each side were recorded. Total area was

computed using calculators which were programmed with a triangulation algorithm.

4.4 Preparation for the Survey

Four enumerators and four enumerators’ assistants were hired. The hiring criteria
were experience in collecting survey data and good pass grades in mathematics, English and
agriculture at 0-levels. Enumerators went through a three-week training period, the propose
of which was to:

a) acquaint them with the overall objectives of the study;

b) explain the role of an enumerator in a survey;

c) explain how to conduct a good interview;

(1) communicate the purpose of each question in the questionnaire;
e) explain how to record responses to questions.

Draft questionnaires were prepared with the help of faculty and colleagues at the
Center for Social Research. The draft questionnaires were used for mock interviews during
the training of enumerators. Pretesting these questionnaires gave an opportunity to gauge the
clarity of concepts in the local language as well as providing more interviewing experience to
the enumerators. Three field tests were done. After each test, ambiguous questions were
reviewed. The major criteria for keeping or revising questions were as follows:
a) Does the question elicit the needed information?
b) Will the respondent understand the question?
c) Will the respondent know the answer?
(1) Will the respondent reveal the answer? (Casley and Lury 1981)

53
Preparing enumerators for measuring gardens took one week of training in reading
bearings and taking land measurements. This was done in a second visit after the household
survey was completed. Although enumerators were very familiar with the arithmetic
processes involved, the physical implementation is a meticulous exercise. A closing gap error

of 5 % was the maximum accepted in making computations.

4.5 The Survey

By December all the instruments for the study were ready. Each household was
visited at least twice. The first time was to complete the household questionnaire, and the
second time to measure the parcels or plots of land. In some cases, measuring the gardens
took more than one day if the household had many plots or if they were far apart. Each
enumerator interviewed 2 to 3 heads of households in a day. The household survey took 40
working days to complete. The measuring of gardens was the most difficult job of the
survey. Torrential rains made it difficult to measure the gardens because of flooding. This
was especially difficult for dimba gardens where in some cases water was knee high. A
second problem was to get the farmers to reveal all of their gardens. Some were suspicious
that enumerators were government agents trying to confiscate their land. Others, by then
convinced that enumerators were not government agents, revealed more plots for which

additional enterprise sheets had to be completed.

4.6 Secondary Data Collection

Secondary data collection was necessary to supplement the data collected through the
farmer survey. This was particularly important to gather information on yields of irrigated
crops and the techniques used for irrigation.

Extensive discussions were held with groups of farmers at Mchenga and Diamphwe

irrigation projects and their extension workers, as well as senior officers in the Department of

54

Irrigation and the Land Husbandry section of Lilongwe ADD. This was done to get
information on:

- crop yields for irrigated crops not currently grown by farmers practicing rainfed

agriculture

- suitable crop combinations for proposed irrigation technologies

- input cost and labor requirements of these crops

- costs and serviceability of some of the irrigation technology

- limitations and expected problems with technologies

- organizational arrangements for multifamily irrigation projects

A visit was made to the pilot sprinkler irrigation projects in Lilongwe district.
Information collected included experience in organizing for the project (e. g. land re-
allocation, set-up logistical problems, marketing of produce, operation of equipment). The
investigator’s experience of supervising a nutrition survey in an irrigated rice growing area,
Bundi, in Karonga district also proved to be beneficial. This gave some perspective on
farmers’ problems in running communal irrigation schemes. This area has both government-
assisted and self-help schemes. In Zomba, there is an irrigated vegetable garden which

offered an opportunity to learn of the tank/furrow irrigation method.

4.7 Summary

This chapter describes the planning and implementation of the survey that provides the
data for this study. Information was collected from households using structured
questionnaires for interviews and measurement of gardens. Sampling of households was
purposive to get households with dimba land.

Secondary data sources were used to supplement the farmer survey mostly with data
for yields of crops not currently grown with irrigation. Other data included types of irrigation

techniques and their costs.

 

CHAPTER 5: CHARACTERISTICS OF SURVEY HOUSEHOLDS

This chapter describes survey households, using variables of immediate concern to the
study. These include size of farms; cropping patterns and yields; use of inputs and hired
labor, food security strategies and willingness to participate and pay for irrigation. They are
extensively compared and contrasted with the findings of other agricultural household studies

conducted in the area.

5.1. Household Characteristics

5.1.1 Social and Demographic Characteristics

The major socio-demographic characteristics of households are presented in Table 5.1.
Among the 270 households interviewed, 53 (19.7%) are female headed. This is below the
nearly 40% average proportion of female headed households in Blantyre ADD. Since the
survey was a census of households with dimba land in the area, this result implies that the
proportion of female headed households with dimba land is less than in the general
population.

Fifty two percent of the heads of households can be considered literate: having four
or more years of formal schooling. The Center for Social Research (1990) reported a
proportion of 48% for this datum, for 1990 in this area. The mean household size is 4.8,
ranging from 1 to 12. The median size is 5, and 75% of households have 6 or less people.
There is an association between size of landholding and size of households. At the mean of

5, medium and large farms have one more person than the small farms.

5.1.2 Size and Distribution of Land Holdings

The distribution of households into the three categories of landholding (small, medium
and large) is 16.3%, 36.3% and 47.4% (Table 5.1). This compares with 57.4%, 31.6%, and

11.0% respectively for Blantyre ADD from the National Sample Survey of Agriculture results

55

 

56

(NSSA 1981). It appears that farmers targeted for this study, those with dimba land, have

larger landholdings.

 

 

 

 

Table 5.1
Socio—Economic Characteristics of Households by Size of Landholding
—
Size of Iandholding(Hectare)
Household Characteristics ,
Small Medium Large All
<0.7ha >0.7- >1.49ha Farms
1.49ha
Sample Size 44 98 128 270
(16.3%) (36.3%) (47.4%) (100%)
Female-headed 9 22 22 53
Male-headed 34 77 105 217
Size of Household(Mean) 4.2 4.7 5.2 4.8
Education(% literate >3 years of school) 62.8% 57.6% 45% 52%
— _

 

 

 

 

 

 

 

 

Source: Survey Data 1993

In Table 5.2 are displayed data on size of farms. The mean size of total land holdings
was 1.7 hectares. Means are 0.46, 1.12, and 2.49 for the three farm sizes. The overall
mean size of rainfed land was 1.5 ha and for dimba land 0.24 ha. Land holdings are usually
fragmented into several parcels of land. The median number of parcels of land was 2,
ranging from 1 to 5. The median number of enterprises was 3. At the mean of 2.9 parcels,
small farms have more fragmented land holdings than medium or large farms, both categories
having 2.4 parcels. But they have fewer enterprises, 2.4 compared 2.7 or 3.2. These
statistics are significantly different at the 5 % level. The production significance of this
fragmentation of land is linked to reducing risk of potential losses from localized
environmental, pest or other adverse conditions, and of exploiting variable eco-niches suited
to different crops (Ellis 1988, Peters and Herrera 1989). Some crops do well in dimba soils,

others on high ground. For example pumpkin plants last longer if grown on extinct anthills.

57

Table 5.2
Size (Hectares) and Distribution of Land Holdings of Households

 

 

Cultivated Area Small Medium Large All
Survey Size 44 98 128 270
Total Land

Mean 0.45 1.12 2.57 1.70

Median 0.45 1.10 2.32 1.45

Std. dev. 0.13 0.22 1.00 1.11
Rainfed Land

Mean 0.40 0.98 2.24 1.51

Median 0.39 0.94 2.00 1.25

Std. dev. 0.17 0.24 0.99 1.07
Dimba Land

Mean 0.13 0.17 0.33 0.24

Median 0.09 0.12 0.21 0.14

Std. dev. 0.12 0.16 0.36 0.29
Rainfed Maize Land

Mean 0.39 0.90 1.58 1.16

Median 0.39 0.89 1.50 1.02

Std. dev. 0.12 0.29 0.74 0.70
No. of parcels (mean) 2.9 2.4 2.4 2.5
No. of enterprises (mean) 2.4 2.7 3.2 2.9

 

Source: Survey Data 1993

These data show that survey households have more land than is the case in the
Blantyre ADD. The larger mean land holding is likely because only households with dimba

were included. Dimba land is prime land which not all households have access to. Only

58

those who have lived in the area for a long time and have familial ties to the chief and clan
heads are likely to have this land, and more of rainfcd land.

However, Blantyre ADD, and particular Chiradzulu district, is an area with the
highest concentration of near landless farmers. For example, 35.2% of all the nation’s
households with less titan 0.7 ha of land are in Blantyre ADD, yet the latter has 24% of all
households in Malawi (Center for Social for Research/NSSA 1988, Malawi Government
1984). It is these farmers who need intensive cultivation methods to make good use of their
limited land.

Medium and large farms have on average one more person than small farms. These
statistics agree with the findings of earlier studies in this part of Malawi. For example,
Chipande (1987) reported that a higher proportion of men in small or near landless
households migrate in search of paid employment. Additionally, a large proportion of female
headed households are characteristically short of one male adult, being either never married or

divorced. Peters and Herrera (1989) also concluded that ’land holds men’, p. 52.

5 .2 Farm Characteristics

5. 2.1 Cropping Patterns

The survey households have a total of 395 rainfed hectares of land and 59 hectares of
dimba land. The most dominant crop mixtures according to frequency and area planted, on
both land types are shown in Tables 5.3 and 5.4. A total of 785 enterprises were surveyed
for the 270 households. Of these, 385 (49%) are rainfed enterprises and 403 (51%) dimba
enterprises. An enterprise was identified by asking farmers to name up to three of the most
important crops grown on a plot of land. There is a wide variety of crop mixtures planted in
the area. For rainfed land, 28 enterprises were identified. However, only 8 were grown on
more than 2% of the plots, and these 8 accounted for 90% of the plots of land. Table 5.3
lists crop mixtures grown on more than 2 % of plots. The most common crop (mixtures) are

maize/beans, peas, maize/beans/pigeon peas, maize alone, and maize/beans/groundnuts for

 

59

Table 5.3
Most Dominant Crop Mixturesa

 

Rainfed Crops % Dimba Crop %
Maize/Beans 32.6 Sugarcane 24. 1
Peas 23.5 Maize 12.2
Maize/Beans/Pigeon peas 10.1 Tomato 9.7
Maize 6 . 7 Cabbage 7 .9
Maize/Beans/Groundnuts 4.4 Chinese cabbage 5.5
Maize/Pigeon peas 3.1 Peas 4.0
Maize/Peas 3.1 Rape 3.0
Maize/Pigeon peas/Groundnuts 2.1 Tomato/peas 2.5

 

a/ grown on more than 2% of plots

rainfed land. Because of the way that enterprises were defined, it is highly probable that the
maize/beans combination has some pigeon peas or groundnuts but farmers did not feel that
these were important enough crops to mention. Also, the maize/peas mixture may actually be
maize followed by peas. Maize based mixtures make up 80% of all rainfed enterprises.

For dimba enterprises, 47 crop mixtures were identified. Eighteen of these were
grown on more than 2% of plots and made up for 89% of the dimba plots of land. In dimba
gardens farmers mostly grow sugarcane, maize, tomatoes, and cabbage. Chinese cabbage is
similar to rape or mustard greens, and these crop mixtures together are grown on 10.6% of
dimba plots. Similarly, tomato mixtures are grown on 15.6% of all dimba plots.

These findings are similar to those of Peters and Herrera (1989) who reported maize,
beans and pigeon peas as the most important crops of the area. Due to small representation
of dimba plots in their sample, sugarcane, tomato, cabbage and other leafy vegetables do not
feature prominently in their results.

Multiple cropping and intercropping are important agricultural practices in the study

area. (Multiple cropping is used to refer to cases where crops are grown sequentially on a

 

60

Table 5.4
Most Important Enterprises by Area (Hectares) Covered

 

 

Enterprise , No. of % Area Mean Min Max
Hectares Covered

Rainfed Area 395 87% 1.51 0.13 5.80

Dimba Area 59 13% 0.24 0.01 2.05

Rainfed Entegprises (5 % and above)

Maize/Beans 141.00 35.7% 1.12 0.19 3.88
Peas 91.40 23.5% 1.00 0.09 3.11
Maize 30.25 7.7% 1.16 0.26 2.75
Maize/Beans/Pigeon Peas 40.50 10.3% 1.06 0.28 2.90
Maize/Beans/Groundnuts 19.62 5.0 % 1 .22 0.23 2.45

Dimba Enterprises (3% and above)

Sugarcane 22.43 38.0% 0.23 0.02 1.49
Maize 5.56 9.4% O. 12 0.04 0.51
Cabbage 4.54 7.7% 0.14 0.01 0.53
Tomato 4.13 7.0% 0.11 0.01 0.51
Maize/Beans 1.95 3.3% 0.32 0.03 0.95

 

Source: Survey Data 1993

piece of land in one cropping season. Intercropping is used to refer to cases where crops are
cultivated concurrently.) Only 32% of dimba enterprises were intercropped and this was 65 %
for rainfed enterprises. Eighty-two percent of all parcels of land were either multiple cropped
or intercropped. Peters and Herrera (1989) reported that 71% of all the plots in their sample
were intercropped. The difference may be due to a low representation of dimba gardens,
which in their sample indicates 13%, compared to 51% in our sample. It is in dimba gardens

that farmers mostly practice monocropping.

61

5.2.2 Estimating Yields

Yields for four harvests prior to the survey were estimated from recall data. They are
reported in Tables 5.5 and 5 .6. A clear pattern emerges. All crops yielded poorly in 1992
because of drought. It is difficult to assess the reliability of yield data because there are very
few crops for which farm based yields exist. Government documents tend to report average
yields for a particular crop. The best are those of maize. The comparison for maize is
shown in Table 5.7.

The yield of beans depends on the plant population and the planting patterns. If
planted in the same hill as maize, the yields are generally lower. The decision to plant on the
same hill or not depends on the variety. Climbing varieties are usually planted in the same
station as maize so that maize plants provide support. Assuming the same planting station,
yields ranging from 41 kg to 235 kg have been reported (Jere 1990). The survey results are
within this range.

5.2.3 Family Labor Available and Use

An estimate was made of total family labor available, and its allocation to farm
enterprises. These estimates are used in formulating enterprise budgets and technical
coefficients that are used for programming.

Family labor available to a household for its farm activities depends on the number of
adults and the amount of time they are available to work in the gardens. Labor allocated to
various enterprise is reported in Table 5.8. It is possible to check these data in the case of
maize-based crop mixtures against ARU data as shown in Table 5.9. The comparison is very
good except in a case of land preparation where the survey results are very high. Jere (1990)
has suggested caution in using these ARU data. Since farm workers were under observation,
they may have worked faster.

The differences between ARU and the survey data may be due to the different method

used to collect them in the two studies. The ARU used observation, while in this study the

 

62

recall method was used. Farmers were asked how many people performed a production
activity, and for how many days and hours per day. Land preparation takes a long time
(mode number of days is 30), and when written records are not kept, farmers may have
reported general estimates. This may also be the case for labor use for irrigation, although
there is no data to compare with. The figures on planting and fertilizing are lower in the
survey results. Despite these differences, the survey results are used in programming without

adjustment.

63

 

Nam“ 52¢...)— E 55050.5 Rea—3:34 8 «2:933 San racism ”023m

BNEESZ u .Ez B
BEES u .E a

28:58 2m 85mm Eon 3:88 8a 8ch 099 :85 682

 

am 1. m3 .1. S l a: t.
w: .1 mm: .1 F: 1. v3 1. Sum SEEM
.1 N» l. we 1. we 1. mm—
--- 3 _ -t ofi t- o2 t- X: 326.580
1. “a l- 3 .1 nm 1. av
.1 cm .1 no t- Q ~ 1. 2. meson
.: an 1. as .1 v5 1.. 3m
t- mam t- SN; 1. mmmJ i an; 332
3:5580\§m\on_a2
an ma 3 9: 2.. 3. :3 no
am an c» Q. mm o9 X: 3: Sum 5&5
me an mv em 3. R. em me
3 co 3 H H _ m3 awn cm 93 83m
SN emm Eh ANN.— Nwm EN; :3 n2.—
mmv woo awe one; go 3A.; o: 2 n; 832
mean ccowmfimsononmaE
3— n3 3m «am can 2e mum m8.—
mM: 3m wwv So; m2. 2 _ 4 amp «MN; 382
am an mu an pm an «N 5.
mm on me ac 3. cm 3. mm 83m
3a in gen :8. b3 35 59 9%
New mum in an" J m3 a2; new mm 1— 382
889332
E2 .5 E2 P...— Pn—Z .E ...Hmz 9E
89 5E 89 $9

 

Ease “seesaw 256 832 a was; :28: use 8...:

Wm 035.

_ _ .

N02 Ens—«2 E cocoa—no.5 Rea—3:}. 9 02:93.3 San— »oiam ”088m

858 E m £5. E
BEBE eoz HEZ B

 

 

BNEETE a
5 ame— 33 ”a." man; :86 $3 $5.
0:; RE 33 «3.2 EX 9%.: RS 82.: asap unease
EN; $3 3; man." :3." 33 Ed $3
9.3 $3: :3 Sea can 25.2 33 98.2 at“: 885
$5 33 £3 33 Sum an." .3... «an;
an? 83 $3 $2: :3 an? :3 33 285
e3... .1 5... .1 a8... .1 S3 1.
as.” $3 82 1. an; .8835
en..— Sm... :3 fine 8..." as.» 3.3 2.5.
83 Sm.» omen E? 3mm 33 SE. 83 0326
E2 .5 E2 E E2 E E2 an new

82 82 82 $2

 

see: recess 85 365 co 8.»; EB: e5 an:
en 23.

65

Table 5.7
Comparing Local Variety Maize Yields (Mean kg per ha) with Other Reports

 

 

Source Yields
Year Fertilized Unfertilized All
Survey Data 1992 579 292 ----
1991 1,051 488 ----
1990 1,100 713 ----
1989 1 ,232 789 ----
Smale 1990 1,200 700 ----
BLADD 1991 -—- --- 726
ARU 1989 1 ,859 ---- —---
GTAPM several --—- ---- 800-1 ,200
Note: BLADD = Blantyre Agricultural Development Division
ARU = Adaptive Research Unit
GTAPM = "Guide to Agricultural Production in Malawi (1992)

Source: Survey Data l993/Guide to Agricultural Production in Malawi 1992

66

 

82 5234 E 5:885 REE—8:32 9 025932 San >955 “088m

cozfimaoi 98:— " moi was ...

 

 

Gm 33 8v SN 23 8&3 59553.“:
as 2 m aim 8m a8 83 09256 08:20
m2 we :3 N3 9: 33 as: £85
83 m3; 9% am om: 29:3
m2 RN 3 E Sm 883%
8» R 93 EN n: as; owsfiu
mm 3 R n: 30E BEE
: E mm SN 3 m3. a:€§8928m§§
we 8 3m 3 8m 30E cooméasefiaz
2 mm § 2 wan 0%:
a E i a: pm Sm 888832
mutant“: mENonm cotawt: 9.633 9555 38m 984 8:98?"—

 

Aflsonéofiom agave wear—905m n80 o. @2822 .823 338m

ad 035.

67

Table 5.9
Comparison of Labor Allocation to Maize/Beans Crop Mixture
(Person-hours)

 

Farm Operation

Ridging
Planting
Fertilizing
Weeding
Harvesting

Total labor required

ARU 1

7O
63
84
170
76

463

Own Survey (Median)

301
37
51

169
95

653

 

1/ Source: Blackie MJ. (1989)

68
5. 2.4 Production Expenses
5. 2. 4.1. Hired Labor Expenses

Farm production need not be constrained by the availability of family labor is casual
labor is available for hire. Forty percent of households in the sample hired some labor and
nearly all of it was used on maize. The mean for only those who hired labor is 351 person
hours, and out of this an average of 312 was used on maize. The total amount hired increases
on average with land holding size. However, when the data are put on a per hectare basis,
the smaller households hire significantly more labor. This may be due to the fact that a large
minority of those who have small land holdings are employed so that they have to hire labor
(see Table 5.12).

On an enterprise basis, the survey results on hired labor (Table 5.10) indicate that it
is mostly used on maize-based fertilized crop mixtures. The total amount of labor hired for
these crop mixtures is shown in Table 5.11. Labor is hired mostly for land preparation and
weeding.

Expenses for hired labor were estimated to be on average MK42.3, based on an
assumed wage rate of MKO.25. If lower wage rates are used7, the estimate may be close to
that reported by Peters and Herrera (1989). They reported that 40% of their sample
households had hired labor, and a modicum of expenses for hired labor: MK2.99, MK8.51,
and MK34.00 for three categories of farmers respectively. Smale (1991) reported an overall
mean hired labor expenses of MK14 per year in BLADD.

5. 2.4.2 Fertilizer Expenses

Eighty-one percent of the households used some fertilizer. The proportion of

households using fertilizer varies from 67.4% in the smallest land holdings to 75.8% in the

 

7 The actual wage rates used are not well established empirically. In this study we used
sensitivity analysis to assess the impact of lower wages on returns to farming activities.

69

medium category to 89% for the largest land holdings. These figures compare well with the
findings of the Center for Social Research of 72% for 1990 in Chiradzulu North EPA.

On average, 147 kg of fertilizer was used on all crops. Most fertilizers cost MK1.00
per kg. The total amount used varies from 25 kg to 1,725 kg and increases with landholding
size. Seventy-eight percent of households used fertilizer on maize in the 1991/92 cropping
season, applying an average of 132 kg. Of the dimba plots, 29% had some fertilizer applied
to them, and this was 55% for rainfed enterprises.

Although many dimba plots were fertilized, only two enterprises were fertilized by
more than 10 farmers: cabbage and chinese cabbage. They applied a mean of 82 kg per ha
and 48 kg per ha of fertilizers respectively. The four major maize-based mixture crops had
fertilizer applied to them in varying quantities. These levels of fertilizer use are reflected in
the yields of the enterprises: maize/beans/pigeon peas having the highest yields and fertilizer

application. See Table 5.12 and refer to Table 5.5.

70

NS: 5252 5 5:52:55 E§3Ew< 9 523333 Sun 53am 583m

 

:

5:55:

5
mm

2
a
8
.2
G
SN
5

mm vw
wv vw
he mg
a:
l- N: 5
1. mg
l- mam
l- v
3 ov
me. o: N
9» com
o hm

55:5": 5:503

l- on
l- 8:
1. am:
.1 C
Om c:
Om vw
mm 2:
v S
1. no:
1. 0mm
.1 mwm
l- m
l- ow
l- ova
l- aNN
3
l- Vm
.1 com
.1 m—N
l- m
C: on
vu m3
mm 5N
w am

5:55 555 :51—
b_>:o< 55255 580

>535
5:52
552
€5505m=m
>25ch
5:52
552
8555 5:55
>5:ch
5:52
552

3555529558352
>5:ch
5:52
552
3555 :oow5\5om\oa_m2
>53
5:52
552
@832
>535
5:52
552
A5 55.55552

3:955 505

 

53:3. 55255 no.0 E omteofim .3 352.cofi5v 85.: “5.55 no 8:

o_.m 25,—.

71

 

 

Table 5.11
Hired Labor (Person-hours) by Cr0p Enterprise

Total Per hectare
Crop Na Mean Std. Dev. Mean Std. Dev.
Maize/Beans 42 497 (422) 286 (302)
Maize 6 374 (324) 203 (167)
Maize/Beans/Pigeon Peas 20 407 (470) 385 (397)
Maize/Beans/Gnuts 7 574 (438) 285 ( 177)
Rainfed Peas 23 147 (107) 73 (64)
Sugarcane 19 203 (291) 101 (103)

 

a! N = number of households who actually hired labor for at least one crop production
activity

Source: Survey Data l993/Guide to Agricultural Production in Malawi 1992

- i

72

Table 5.12
Use of Fertilizer kg/ha by Crop Enterprise and Size of Farm

 

 

Small Medium Large All

Maize/Beans

#‘ Applying 8 37 63 108

Mean 119 155 157 154

Std. Dev. 117 103 86 94
Maize

# Applying 1 5 11 17

Mean 100 108 157 138

Std. Dev. --- 68 133 113
Maize/Beans/P. Peas

# Applying 4 13 12 29

Mean 175 288 198 235

Std. Dev. 96 445 147 311
Maize/Beans/Gnuts

# Applying 4 3 8 15

Mean 131 192 188 173

Std. Dev. 63 52 88 76
Cabbage ‘

# Applying 3 10 6 19

Mean 187 69 54 82

Std. Dev. 271 50 62 113
Chinese Cabbage

# Applying 9 3 5 10

Mean 44 27 63 48

Std. Dev. ? 22 48 38

l/ I! = number

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

73
5. 2. 4. 3 Pesticides Expenses

Many farmers said pests were a major constraint to growing dimba crops. One-third
of farmers used pesticides in small quantities on 99 (12.6% of total) crop enterprises. They
were applied mostly to dimba crops: cabbage, chinese cabbage, tomato, and rape. The most
commonly used pesticides were DDT (43.4%), Actellic (36.45), and Malathion (10%). Other
types included Copper, Sevin, Diamethoate 45, and Ripcord. The expenses are shown in

current enterprise budgets in Tables in Appendix A.

5.3. Income Generating Activities and Off-Own Farm Employment

Income generating and off-farm employment activities are important sources of
income for many smallholder families. In this study they are also important because they can
provide competition for irrigated cultivation, with either positive or negative effects. Income
generating activities (IGAs) are self employment jobs in trades like commodity trading, brick-
making or basket weaving. (In later chapters another term, income earning activities (IRA),
is introduced which refers to all non-farm income generating activities including wage
employment.)

Farming their own gardens is considered the main occupation8 by 71.5% of the heads
of households. Nearly 6% were employed outside the agricultural sector, 2.2% were
agricultural workers, and 6% did some IGAs. Overall, 13.8% said that farming their own
gardens was not a primary source of livelihood. lGAs were part of or the only source of
livelihood for 11.5% for the households. The Center for Social Research (1990) had similar
findings; 72.9% of the heads of households relied on farming. However, a higher proportion,
16% , were in non-agricultural employment.

The importance of off-farm incomes varies inversely with size of land holding. Twice
the proportion of households (27%) in the smallest landholding category as in the larger two

said farming was not a main occupation. The results tally with findings of other studies

 

3 OcCupation was defined as main source of livelihood.

74
which indicate negative relationship between land holding sizes and non-farm income
generating activities.

A shortage of land is associated with the allocation of men’s time to off farm
activities. Men in the smallest land holdings (under one hectare) allocate far more of
their time to retailing (fish or agricultural products), and wage employment (Peters
and Herrera 1989, p. 52).

Peters and Herrera (1989) also reported that on average 40% of their sample
households income came from off-farm sources. Chipande (1987) conducted a study in
Phalombe, a project adjacent to the one where this study was conducted. He found a greater
tendency for men in households with smaller landholding to seek paid employment on estates
or the nearby urban centers.

The preponderance of off-farm sources of income in this area reflects the
configuration of opportunities available to earn a living. There have been quotas or
prohibitions on growing some high value crops such as tobacco. This, together with low
gross margins of crops currently grown and proximity to a major urban center of Limbe—
Blantyre, have made off-farm sources of income very attractive. This is in contrast to the
situation in other parts of Malawi. Farmers’ cash-generating activities are mostly selling
livestock, selling non-maize crops, and remittances from away household members (Smale
1990, p. 5, Center for Social Research 1988, p. 121).

Thus farmers in the study area could be termed part-time farmers. The principal
objective for many of these farmers is to assure a relatively cheap food supply, while
generating income from off-farm activities (Smale 1990 p. 6). It may also be the case that
farming is underwritten by incoming transfers, either from members of the family working
elsewhere on a permanent basis or living in the same household.

Seventy-one percent of household heads said they had some slack months in the
agricultural calendar, and during these months 44 % of these engaged in some kind of income
generating activity. No detailed data was collected on nature of IGAs except incomes made.
These ranged from MKS to MKSOO per month, with a mean of MK101. These figures are

very high but since they are seasonal, per annum earnings are very low.

75

Table 5.13
Main Occupation of Household Heads

 

 

Main Occupation Percent
Farming 71.5
Non—Agricultural Employment 5.6
Agricultural Worker 2.2
Income Generating Activities 6.0
Farming and Ganyu 4.5
Farming and Non-Agric. Employment 2.6
Farming and IGAs 5.9

All Occupations 100.0

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

Table 5.14
Proportion of Farm Households Engaged in Non-Farm Work by Farm Size

 

 

Small Medium Large All Farms
Engaged in IGAs 14% 9.1% 12.6% 11.5%
Off-Own Farm Occupation 27% 12% 14% 16.7%

 

Source: Survey Data l993/Guide to Agricultural Production in Malawi 1992

76
5.4 Maize Production and Food Security

In this section data are presented and discussed with respect to food security situation

of households. The emphasis is on strategies to acquire food.

5. 4. 1 Maize Harvests

Mean total rainfed maize harvests for the four years prior to the survey (1988/89,
1989/90, 1990/91, and 1991/92) are reported in Table 5.15. They were 1,021, 1,031, 928,
and 463 kilograms respectively. We would expect these to be generally higher than the i
harvests in the general smallholder farming population are because most of our sample used 5‘
fertilizer and have larger land holdings. For example, in 1990, the Center for Social
Research estimated mean rainfed maize harvests of 476 kg. The sample had more households
with smaller farms. In addition, 1990 was a year of devastating floods. Peters and Herrera
(1989) estimated an average maize harvest of 882 kg for 1986/87 cropping season. The size
of farms in their study are comparable to this one.

On average the survey households are maize self-sufficient but the drought easily
changes this situation. In 1991/92, the maize produced per person was below the level of
200 kg in all the three categories of households. Fertilized maize yielded twice as much
unfertilized maize. Female-headed households harvested 615 kg in 1989/91 and 293 in
1990/91 compared to 982 kg and 505 kg for male-headed households in the same years.

Thus, female-headed households produced on average 60 % of what male-headed households
did. The differences in total harvests by land size category and gender of head were

statistically significant at the 5% level.

77

Table 5.15
Mean Rainfed Maize Harvests (Kilograms) by Size of Farm in 1989 - 1992

 

 

1988/89 1989/90 1090/91 1991/92
Small 681 (I80)' 695 (208) 630 (I80) 247 (69)
Medium 873 (265) 930 (269) 866 (252) 441 (139)
Large 1228 (296) 1206 (274) 1064 (238) 544 (123)
All households 1021 (269) 1031 (361) 928 (235) 463 (122)

 

a/ Italicized figures are per capita

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

Table 5.16
Mean Harvests of Fertilized and Unfertilized Maize (Kilograms)
by Size of Farm in 1991/92

 

 

 

Small Medium Large Total
Fertilized 256 51 l 562 508
Maize 68 I60 123 129
Unfertilized 229 246 330 268
Maize 69 84 103 85
All Maize 247 441 544 463

69 139 123 121
a/ Italicized figures are per capita

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

11“” V

78
The food security position of households can be established partly by how long it
takes to exhaust their maize harvests. The households were grouped into those who deplete
food stocks before harvest and those who do not. The proportion of households whose food
stocks lasted from one harvest to another varied in the three years prior to the survey: 0.7%
for 1991/92, 11.9% for 1990/91, and 48% for 1989/90. These figure were collected through
recall data by farmers and should be treated with caution. However, they bring out the

general trend of poor harvests in 1991/92 due to drought.

5. 4.2 Food Security and Income Strategies

Only 31% said their food security strategy was to produce all food requirements.
Thirty-eight percent grow cash crops in addition to growing food to be food secure. A good
7.4 % of households relied on "ganyu”9 labor to be food secure, and an additional 5.3 %
complemented other strategies with ganyu labor. Seven percent did some kind of IGA.
Nearly 10% combined own farm production with other strategies (see Table 5.17). The
major differences in strategy by farm size are that fewer small and medium size farm
households depend only on farm production, and more small farms rely on "ganyu" for food
security.

For the forty five percent of those who rely on agricultural cultivation to be food
secure, dimba crop production was the major way of managing yield risk in case of rain
failure. Trading in agricultural commodities was an important alternative strategy for another

15.5%.

 

9 Ganyu is piece work wage labor. It is an important survival strategy for going through
the hunger months of December to February (Center for Social research 1990), and is also a
way of getting seed (Smale 1990).

79

Table 5.17
Food Security and Income Strategies by Farm Size (Percent of Sample)

 

All Small Medium Large

1. Produce All Food 31.1 30.2 37.4 41.4
2. Produce Food and Cash Crops 38.1 30.2 26.3 35.2
3. Own Farm Production + Others 9.6 4.6 6.1 4.0
4. ”Ganyu" 7.4 14.0 5.1 7.0
5. Employed Permanently 3.0 4.7 5.1 1.0
6. Income Generating Activities 6.3 9.3 7 .1 4.7
7. Others 4.5 10.0 12.1 6.6

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

5.5 Farmers’ Willingness to Use Irrigation

Willingness to invest in formal irrigation should be related to perception of constraints
to rainfed and dimba farming. For rainfed land, farmers felt that lack of fertilizers (31%) and
poor rainfall (27%) were the major constraints. Poor rainfall and water shortage were also
important constraints for dimba land cultivation (21% and 14.9% respectively). However,
pests and fertilizer were also mentioned frequently (Table 5.18).

Seventy-one percent said they would be willing to participate in irrigation. Sixty-six
percent foresaw problems in organization and implementation. Examples of this are disputes
among farmers and fear of getting involved in government credit. The crops they frequently
mentioned as wanting to grow were maize, tomatoes, and cabbage. The single most

important reason for choosing the crops was to generate cash.

80
Table 5.18
Constraints to Producing More Crops

Percent of Farmers
Mentioning Constraint

 

Constraint

Rainfed Land Dimba Land
Lack of Fertilizer 31 13
Poor Rainfall 27 22
Water Shortage -- 15
Pests 0 15
Others 42 34

 

Source: Survey Data l993/Guide to Agricultural Production in Malawi 1992

5.6 Summary

This chapter presented the major descriptive characteristics of the survey households.
The number of households surveyed was 270. of which nearly 20% are female headed. Fifty-
tvo percent of heads of households are literate. Median household size is 5. The households
are generally better endowed with land than most households in the area, with a mean land
holding size of 1.7 ha, of which 1.5 is rainfed and 0.24 ha is dimba land; 16.3% are under
0.7 ha, 36.3% between 0.7 and 1.49, and 47.4% greater than 1.49 ha.

The dominant crop mixtures grown on rainfed gardens are maize/beans, peas, maize,
maize/beans/pigeon peas, and maize/beans/groundnuts. On dimba gardens these are
Sugarcane, maize, cabbage, and tomato.

The households are more likely to use fertilizer than is the case among typical
Smallholder farmers in Malawi. A large minority hire labor for fertilizer maize-based crop
l‘l‘lixtures grown on rainfed land. Pesticides are used on dimba crops of chinese cabbage,
Cabbage, tomato and rape. 9

Although all households farm some land, nearly 30% do not consider farming their

major occupation. Even in good rainfall years many households do not produce enough

81
maize. The majority, 71%, are willing to participate in irrigation. However, they foresee

problems in organization and implementation.

cm 6: ENTERPRISE BUDGETS AND FORMULATION OF THE TARGET
MOTAD MODEL

This chapter describes how the safety-first whole-farm Target MOTAD programming
model was formulated, using enterprise budgets and other data. The emphasis is on the
sources of data, reasons for including specific activities, and the derivation of coefficients in

the programming matrix.

6.1 Enterprise Budgets

An enterprise budget is prepared by stating income, expenses, and resource needs of
the enterprise on a per unit basis (Harsh et al 1981). Crop enterprise budgets in this study are
constructed for two reasons: a) to calculate gross margins for each enterprise; and b) to
calculate the technical co—efficients needed for the programming matrix presented in section
6.2. Incomes or gross values are derived by multiplying average yields for the enterprises by
the price(s) of the output(s) and subtracting variable costs. Technical co-efficients are
comprised of the per unit input requirements for the production activities of crops in the
enterprises.

The budgets are formulated for currently grown crops, intensive and irrigated crop
enterprises to fit the three models that are used. Model I deals with the optimization of the
representative farm’s returns with current cropping practices. Model II represents the
oPpor’tunities available for farmers to upgrade their cropping practices to intensive status by
using more chemical inputs and the requisite labor. Model 111 allows representative farms to
S“fleet crops grown with current or intensive practices on rainfed land, and replaces dimba
crops with irrigated ones. Model III has two variations. In Model III(a) the irrigated crops
are cereals, while in Model III(b) they are horticultural crops.

Current crop enterprises are those that farmers are practicing in their gardens, as

l'eflec:ted by the survey results. They are of two types: those that use fertilizer and those that

do not. Intensive crop enterprises are similar to the fertilized options of current enterprises

82

83
but the fertilizer and pesticides levels are raised to the recommended levels. Irrigated crop
enterprises are those grown by farmers on irrigation projects. They use recommended
fertilizer and chemical input levels.
The enterprise budgets are presented in Tables A61 to A6.40 in Appendix B. The

chapter discusses how they were compiled.

6.1. 1. Estimating Yields
Yield data came from the survey results, the Guide to Agricultural Production in
Malawi-1992 (GTAPM) and Adaptive Research Unit (ARU) reports. The source of data for

each particular enterprise depended on whether it was current, intensive or irrigated.

6. 1 . 1 . 1 Current Crop Enterprises

Only those crop mixes grown by 10 or more farmers in the sample were considered
for enterprise budgets and eventual inclusion in the programming matrix as crop production
activities. This was done so that the solution carries crops that reflect dominant farmer
Practices and facilitates validation of the model. For each crop mixture there were potentially
tWO options: the first Option where farmers used no fertilizer and the second option where
fertilizer was used. Use of some amount of fertilizer, even if less than the recommended
1eVel, was interpreted as use of fertilizer. This was done to capture the production function
effect of fertilizer use.

The yield data used are typical values for the cases in the sample. For rainfed
enterprises, the mean was used. These values were fairly comparable to other survey results
on Yields. In the case of dimba crops, it is prudent to use the median yields. Due to the
Small size of dimba plots and lack of experience of farmers in reporting harvests of dimba

Creps , the mean could embody serious statistical biases.

84
Based on the above criterion, fourteen crop mixes were selected from the observed
99. Six are rainfed enterprises and the rest are dimba enterprises. When the fertilized and

non-fertilized options are allowed, 15 current enterprise budgets were compiled. These are:

Enterprise Name Abbreviation
Fertilized Maize/Beans ManF
Non-fertilized Maize/Beans ManN
Fertilized Maize MzF
Non-fertilized Maize MzN
Fertilized Maize/Beans/P.peas MzBPpF
Non-fertilized Maize/Beans/P.peas MzBPpN
Fertilized Maize/Beans/Gnuts MzBGnF
Rainfed Peas Rpeas
Sugarcane Scane
Dimba Peas Dpeas
Non-fertilized Cabbage CabN
Non-fertilized Chinese Cabbage CcabN

N on-Fertilized Tomato TomN
Dimba Maize szn
Dimba Maize/Pumpkin Leaves DMzPLN

6. 1 . 1 .2 Intensive Crop Enterprises

Intensive enterprises differ from current ones in using fertilizer at the higher
1‘ eCOmmended levels. This leads to higher yields, and consequently, more labor is required
for harvesting. These levels are not what the survey farmers use, so other sources of data
Were used. The first was the ARU for yields of crops. For enterprises on which the ARU
does not have yield data, those from the GTAPM were used. The following intensive crop

enterprises were included:

85

Hybrid Maize HMaize
Hybrid Maize/Beans/Pegion Peas HMzBP
Maize/Beans Imzbn
Maize/Beans/P.peas Imszp
Maize/Beans/Groundnuts Imzbgn
Cabbage Icab
Chinese Cabbage Iccab
Tomato ltom
Maize M2

6. 1. 1.3 Irrigated Crop Enterprises

Only the fertilized option was specified for irrigated crop enterprises. Twenty- two

 

irrigated crop enterprise budgets were compiled, viz:

Tomato RToml , RTom2, Rtom3
Beans RBeansl, RBeansZ, RBeans3

Cabbage RCabl, RCab2, RCab3
Chinese Cabbage RCcabl , RCcab2, RCcab3
Pumpkin Leaves RPLvsl , RPLvs2, RPLvs3
Onions ROn2

Rice RRice1,RRice2

Hybrid Maize RHymzl, RHymzZ

Wheat RWheatl , RWheat2

The suffixed figures refer to the seasons in which the crop is grown. An assumption is made
that vegetable crops can be grown and harvested in four months. Three seasons are specified
for Model III(a): Season 1 is from January to April, Season 2 from May to August and
Season 3 from September to December. Consequently, enterprise budgets for irrigated crops
are formed with one-third of amortized debt apportioned to each of the three seasons. For
Model III(b)-with irrigated cereals—two seasons are specified: September to March, the
rainy cropping season; and April to August, the winter cropping season. Half 'of amortized
debt is used.

The yields are calculated using information from discussions with farmers at irrigation
proj ects and reports from Lilongwe ADD for Mchenga and Diamphwe irrigation projects.

They are shown in Table 6.1.

86

Table 6.1
Yields of Irrigated Crop Enterprises

 

Seasons 1 & 3 Season 2
Crop Area Area
Enterprise Planted (MP Yield (kg) Planted (Lia) Yield (kg)
Tomato 1 .008 2491 0. 504 5243
Onion --- -- 0.504 8008
Beans 1.240 3738 1.008 4754
Chinese Cab. 0.504 3427 0.504 12001
Cabbage 1.008 51 10 0.756 24312
Pumpkin Lv. 0.252 1670 ~---- -----

 

1/ Area planted refers the area sown to the crop at Mchenga irrigation project in the

respective seasons.

Note: Except for chinese cabbage and cabbage, all the yields are below half of
potential.

Source: LADD 1993(b)

6.1.2 Valuing Yields and Inputs

Yields and inputs were valued using prices from the 1990/91 cropping season for two
reasons:

a) It is the midpoint between 1989 and 1992 so that assuming constant inflation rate we
do not have to deflate/inflate the values; and
b) The yield data available are for 1992. Assuming lagged adaptive expectations, the

1990/91 prices were expected when planting crops in the 1991/92 season. These then

were the prices entering the derived demand for inputs in that year. Also important is

that this was a year of no drought.

If output is sold, the appropriate price for valuing yields is clearly the sale price. But
all farmers in this study consumed some of their output. The consumed output can be valued
at its opportunity cost which is the price the output could have been sold at had it not been

c"3113mmed. This would have to be the average price for the period over which the crop was

consumed, assuming comtant rate of consumption. However, no data were collected on

 

87
quantities consumed. Therefore, the prices used were selected from those of the survey and
those paid by ADMARC in the 1990/91 cropping season. The approach to off-farm income

earning activities is assumed that net returns per hour are equal to the minimum wage rate.

6. 1. 3 Variable Costs

Variable costs include costs of hired labor, seed, marketing, inputs such as fertilizers
and chemical inputs, equipment repairs and maintenance, and operating interest charge.
Excluded from this are taxes, depreciation, insurance and overhead labor costs. The approach
taken concerning interest charges is to include them as part of investment cost in amortized

debt .

6.1.3.1 Seeds

No data were collected through the survey on the seed rate for the enterprises. The
data used are from the GTAPM, which specifies different seed rates for different crops
depending on the mixtures of crops. The seed rate appears in each enterprise budget under

the row title ’seed’. For example, the seed rate for maize is 25 kg per ha.

6- I . 3.2 Fertilizer and Chemicals Applied to Crops

The figures used for fertilizer and chemicals applied to current enterprises are
averages of quantities used by survey households. For intensive enterprises, GTAPM rates
Were used, and the same data were applied to irrigated enterprises. For intensive and
irrigated enterprises, an assumption was made that pesticides and fungicides are applied twice
at the specified rate. Fertilizers are mostly used on maize. The most commonly used types
Were 2020.0 (17%), Calcium Ammonium Nitrate (32 %), Urea (25%), Double Ammonium

Phosphate (4.8%) and 2321.0 (8.7%). The recommended practice is to apply fertilizers
twice, basal and top dressing. Among the surveyed households who used fertilizer on maize,

only half followed basal with top dressing. Top, dressing was done with mostly CAN and

 

88
Urea. Fertilizer costs on average MK1.00 per kg. The quantity of dry fertilizer
recommended for the various crops is reflected in the enterprise budgets. For maize, these
are 370 kg for hybrid varieties and 160 kg for local maize. Among vegetables, an example
can be used of tomatoes. It is recommended to apply every six weeks a total of 80 kg of
fertilizer. Chemicals recommended are Carbaryl 85 WP and Dimethoate 40 WP for aphids,

both applied by spraying solution of 35 g in 14 liters of water.

6. 1 . 3.3 Repairs and Maintenance
No major farm implements are used in this part of Malawi. The major repair and
maintenance cost for any of the activities is that of hoes and Slashers. The exception is for

irrigated enterprises, and this is discussed under Section 6.1.5.

6.1.4 Labor Use and Costs
6. I- 4.1 Family Labor

Family labor does not enter the calculation of gross margin as a variable cost but is
used to calculate returns per unit of labor and as a constraint in programming. For current
enterprises, the median of labor person-hours applied to each enterprise was used. The
exeeption is for land preparation where the median is used for the whole sample of dimba
Crops, excluding sugarcane. This was done for consistency. Time taken to prepare land
ShOuld not vary much by enterprise except for sugarcane, which is a perennial crop and has
Very different planting practices.

For programming purposes, the labor requirements were distributed into the main
mOnths in which the activities for the enterprise were done. This is meant to increase the
Precision of resource constraints in programming. Also, the cost of labor may not be constant

over the year. This approach also expedites sensitivity analysis.

In the case of intensive enterprises, the levels of family labor used are derived this

Way, For the basic activities whose labor requirements do not vary with outputs or inputs

 

89

(i.e. land preparation, planting, weeding, and irrigation), the labor requirement used is the
same as for fertilized current activities. But for fertilizing and harvesting, the figures for
current enterprises are increased by the proportionate increase in harvests or inputs due to
intensive use of inputs. For irrigated crops, the estimates are done similarly to intensively
cultivated crops except that the labor requirements for irrigation are calculated from reports
by the Lilongwe ADD for the irrigation projects. Setting up equipment requires at least two
people, and it takes 2 to 3 hours to irrigate 3 hectare of land.

Some of the labor requirements reported by farmers are very high, especially for
irrigating dimba crops. Current dimba enterprises require more labor for irrigation than
irrigated enterprises because watering is done manually with watering cans. However, there

is need for empirical studies to verify these figures.

6.1.4.2 Hired Labor

Hired labor is used in crop production when it is profitable to do so but family labor
is limiting. The minimum wage is used as rate of remuneration. This is debatable in the case
for "ganyu" labor. The wages are usually in-kind, such as grain, and it is widely believed
that these are very low. However, these views may overlook some costs and benefits implicit
in the wage. First, when grain markets fail, it is attractive for the "ganyu" workers to earn
grain rather than cash because it saves them time and extra money needed to go in search for
grain. Second, the wage should reflect the cost to the employing farmer of storing or
procuring grain and bringing to close to the "ganyu" workers. In the absence of empirical

data, sensitivity analysis is used to assess impact of conjectured wages on returns.

6.1.4.3 Marketing Costs
Marketing costs are calculated as the value of labor used in taking the produce to
market and selling it. The figures are based on the researcher’s experience and observance of

marketing in rural areas of Malawi. People carry produce to the market in the morning and

 

90

expect to complete selling by mid-day. Considering that the maximum load a person can
carry is 50 kg, this works out to about 10 kg per hour. Using the minimum wage of MKO.25
this gives MK0.025 as cost of marketing a kilogram of produce. This figure is multiplied by
the quantity harvestedlo of each crop in an enterprise to arrive at marketing costs per
enterprise”. The exception is sugarcane where it is assumed that each bundle of sugarcane
(20 canes) takes two hours to sell. Thus, a bundle costs MKO.50 and a cane MK0.025 to

sell.

6. 1. 5. Irrigation Costs

Irrigation costs vary by type of irrigation technique used. The base model uses the
costs for sprinkler irrigation. The costs for the various types of irrigation were calculated as
follows:. The estimate is done for five hectares of land because this was the size of the
sprinkler and borehole/furrow irrigation projects the Ministry of Agriculture has implemented.

Sprinkler irrigation equipment cost MK90,000 [LADD 1993(a)]. Figures on costs of
drilling boreholes and installing pumps for other types of irrigation were available from
Ministry of Works, Department of Rural Water Supply. Costs of constructing earthworks
were derived from the data that the Department of Irrigation used to evaluate gravity-fed
projects. This includes the costs of measuring and constructing headworks, canal structures,
road access culverts, bridges, transport, tools, siphons, and payments to surveyors. The costs
per hectare for different sites were estimated from MKSOO to MK3500 depending on ease of
water abstraction and topography. The median, MK2000, is used as an estimate.

Amortized debt is calculated based on 25 % interest rate and 15 years of repayment

period. This period was chosen to coincide with the useful life time of the equipment or

 

‘0 Ideally only the quantity sold should be used. For crops that are mostly consumed e.g.
maize, the marketing costs are a very small proportion, because of low yields. Thus there
should be very little upward bias in the figures.

" There are petty traders who continually sell produce at these markets, but they are not
producers. They can be considered retailers.

91

investment. The high interest rate reflects the current commercial rate of about 21% and
costs of administration and also reflects risks involved in lending over the long time period.

Five types of irrigation techniques were identified, and Table 6.2 below indicates their
costs, amortized costs and annual maintenance costs. The latter are based on 5 % of the value
of investment. For the gravity-fed irrigation, maintenance costs are only for labor used in

keeping the earthworks in good shape.

Table 6.2
Costs of Various Types of Irrigation.

 

Type of Irrigation Investment Cost Amortized Cost Maintenance
Cost

 

Malawi Kwachas

Pump/Sprinkler 18,000 4,664 900
Pump/Furrows 10,000 2,591 500
Bore/Furrows‘ 7,000 1,814 350
Gravity-fed/Furrows 2,000 518 100
Tank/Furrows 6,000 1 ,554 300

 

1/ There are two types of boreholes commonly used in Malawi: the Barnaby and
Afridev. The costs of the Afridev are used in the budgets.

Source: Departments of Rural Water Supply and Irrigation

6.2 Formulation of the Mathematical Programming Model

The model used to determine optimal income earning activities is the Target MOTAD.
It uses Telser’s safety first criterion to model farmers’ risk avoidance behavior. Safety first
goals are specified using two different approaches: minimum maize production and
subsistence income requirements. Three versions of the model are used which correspond
with the crop production practices that farmers can use, viz current practices, intensive

practices and irrigated cultivation.

92

The method used selects an activity mix which optimizes the choice variable, expected
income, while simultaneously enforcing chance constraints up the yearly deviations of income
from the expected optimal income. This is done by using a sufficiency constraint that
imposes Telser’s safety first probablistic criterion in the LP program. The sum of yearly
deviations of income from expected income multiplied by the probability of that year’s state
of nature occurring are summed up. The model then selects a target from which the sum of
probability weighted deviations is subtracted, and the difference is greater than the farmer’s
8031 [t - L*Q(t) > g]-

This section present the programming tableaus used in the models. For reasons of

space the matrix developed is displayed by model showing matrix quadrants with major

activity types.

6. 2.1 The Objective Function

The objective function maximizes gross margin from farm activities and earnings from
off-farm income generating activities subject to constraints on input, market and permissible
chance of failing to meet target objectives. The gross margin corresponding to the optimal
solution is roughly equivalent to net farm income over fixed costs. In the objective function
row are entered the costs or revenues of engaging in a unit of activity (i.e. producing crops,

selling crops, hiring labor or selling labor).

6.2.2 Activities in the Model

The activities in the model include crop production, labor hiring and selling, crop

selling, deviations and chance counters.

cultivai
enterpr
enterpr
objectix
enterpri
efficient

quadran

0f accoux
“10th s
constrain

JANHL 1

Cllllnat 1' C"

”3° year.

of the 3'6;

 

margm - i
for Off‘fal

di3p1336d

 

 

93
6. 2. 2.] Crop Production Activities

A cropping activity is a single variable representing a given crop enterprise that is
cultivated. The unit of account is the hectare. Crop production activities include crop
enterprises grown on rainfed land and dimba land for Model 1. Model 11 adds on intensive
enterprises. Model 111 adds on to Model 11 the crops that are grown with irrigation. In the
objective function row is entered the cost of producing one hectare of the crops in the
enterprise. These are the variable costs in the enterprise budgets. The input-output co-
efficients represent the resource requirements per hectare of the crop activities. The relevant

quadrants of crop production activities are displayed in Tables 6.3 to 6.7.

6. 2. 2.2 Labor Hiring Activities

The model is set up to allow for the hiring of labor if it is profitable to do so. The unit
of account is person-hours. The labor hiring activity for each enterprise is disaggregated into
months so that the optimal solution hires labor only for the month for which there is a
constraint and also because its costs change over the year. Labor hiring activities are labelled

JANHL for January hired labor to DECHL for December hired labor.

6. 2.2.3 Labor Selling Activities

Due to the seasonality of production, especially when only rainfed and dimba
cultivation is practiced, the farm family may have agricultural slack labor in some months of
the year. The model allows for off-farm income earning activities (IEAs) in all the 12 months
of the year. The official minimum wage rate of MKO.25 per person/hour is used as gross
margin. The activities are labelled OFEJAN for off-farm employment in January to OFEDEC
for off-farm employment in December. The quadrants for labor hiring and sale activities are
displayed in Tables 6.8 and 6.9.

6.2.2

ofcro;
the crc

anus

ran—4g¢_1AI

CTOp S

6.2.2.
manix
DYEA
aCls 18

6.2. 3

Constra

 

6. 2.2.4 Crop Selling Activities

94

Crop selling activities together with the yield accounting rows allow for the translating

of crop yields into gross margins. The objective function rows carry the relevant prices for

the crop. The base prices are those of the 1990/91 cropping season. The following sell

activities are included:

Sell Rainfed Maize
Sell Dimba Maize
Sell Rainfed Beans
Sell Pigeon Peas

Sell Groundnuts

Sell Rainfed Peas
Sell Dimba Peas

Sell Sugarcane

Sell Pumpkin Leaves
Sell Tomato

Sell Tomato Seasons 1 and 3
Sell Chinese Cabbage
Sell Cabbage

Sell Irrigated Beans
Sell Onions

Buy Maize

Sell Wheat

Sell Rice

SRMZ
SDMZ
SBNS
SP/PEA
SG/ NUT
SRPEA
SDPEA
SSCANE
SP/LV
STOM
STOM1&3
SCCAB
SCAB
SIRBN S
SIRON2
BUYMZ
SWHEAT
SRICE

Crop sale activities quadrant is shown in Table 6.10.

6. 2.2.5 Deviations and Chance Counters

Since the model is a Target MOTAD, deviations and chance rows are used in the

matrix. Four deviation rows are included to represent the four years of yield data, DYEARI,

DYEAR2, DYEAR3 and DYEAR4; the TQ(t) column sums the deviations, and the t column

acts as a counter that selects the target. The chance quadrant is in Table 6.11.

6.2.3 Constraints

The model maximizes gross margin subject to family labor, land and chance

constraints. We had intended to use a rotation constraint. Several methods are available to

95

model them (El-Nazer et al 1986, Musser et al 1985). But lack of data on crop yields grown
in rotation made this impossible. Also due to lack of data for yields of onions in Season 1

and 3, only Season 2 onions are modelled.

6. 2. 3.1 Family Labor

The average size of household was used to arrive at the available person-hours. An
equivalent of 2 adults available to work full time per household is assumed. They work an
average of six hours a day, 25 days in a month. This is the labor available for all economic
activities be they farm or non-farm. The family labor available is varied by month to reflect
seasonal variation in hours that can be worked due to rains and also disease. In the rainy
season the incidence of diarrhoea and malaria is higher. So in November, December,
January, and February about 5 work hours per day are specified. This gives 240 hours per
household in these months compared to 300 in the other eight months. In the matrix, the
monthly labor requirements are labeled J ANLAB through DECLAB respectively.

6.2.3.2 Land

Land available for cultivation is limited. The model allows for three size pairs of land
holdings for rainfed and dimba land. These are (0.45, 0.13), (1.06, 0.17), and (2.49, 0.33)
hectares for small, medium and large farms respectively. In the case of irrigation, the land
sizes specified correspond to the current dimba land: 0.13 ha, 0.17 ha and 0.33 ha. Irrigated
land available is modelled in seasons for ease of programming crops grown in different
seasons. Three seasons are modelled for production of horticultural crops Model III(a), and

two seasons for cereal crops Model III(b). The mnemonic names used are:

Rainfed Land RFLAN D
Relay Use of Rainfed Land RFLD2
Dimba Land DLAN D
Irrigated Land Season 1 IRLDSI
Irrigated Land Season 2 IRLDS2

Irrigated Land Season 3 IRLDSB

96
Farmers in Malawi use rainfed land to grow field peas or beans or sweet potatoes in
relay with maize. Among the survey households the most common relay crop is field peas.

They are planted around February and March when maize is beginning to dry.

97

 

 

 

05 5 5- web- 355- 3m- 58. van- 2505. mg

So

200me

5 N05

wean- m<0

wcw5- _ @400

05 R. 5.20.5.

5am- w>s5n5

meg- m2<0m

n 5 5.- 3.50

wn 5- and

55.5. ”5.320

mm- 5a. 3.5

3.. cm. :5 - 2.- m2<mm

one 5 - omen- an- MN5<2Q

mm: «mm can whom a: $9. who 2. «N5 nn5 em Nv 5N- cm 3. 023%

h~5~ mam 55c 32 Row «new. Rem vm5 one men new 255 85 35 mg 025m>

NMMN 53 v9. anew new 39 coon $5 59. 5m 5m SN 55w 2.5 nvn 025N>

comm n35 5.5555 onnm name nmun omen 35 can nnN nov wNN 5mm 5o5 3a 0255?

35 5.55 mm 055 mm 555 mm 555 30mm

85 85 en e5 e5 Q5 Q5 e5 95 a5 :02

5.5.5 o5 Q5 05 .55 e5 a5 m5 55.5.00

new new 85 an 3N 5n5 5M5 5m5 5W5 5n5 5n5 5n5 555mm

3N5 o~5 3N5 35 EN now am 5n5 5W5 5M5 5w5 5m5 5m5 5W5 Q5055<

n5n new 0: m5m Nun. $5. an 335.

mnn own 52. new £25.55.

can new 8: 3m 5 we an an we 3 an R we 32

m5n m5n m5n m5n 35 an an we mm an R xv 555%

E. 352

min—mm

«55 on 35 5.55 mm $55 mm 555 mu 555 32¢:

5 5 5 5 5 5 5 925n—

5 5 5 5 5 5 5 n52<5n5~5

be- 55.. en- 8. 95. can- 85- 551 man. Nb- can- 5N- 5n 5- mm- 53- 253.85
2.520 2.580 555 229 25.5.5. 950% 250 50.55 55mg 2552 "5.5; 22 .52 2:555 “5:552

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.9.885. 86 2.55 2a BEE .50 - 5 .33:

m6 mus—5:5.

 

98

 

coma-

«w—w
vnhcp
momo
nmnw

amp
o~w
-k
sen
coc-
m—m

no.c<m-

nouom-

vnmop
owner
0¢ho5
nm¢n

Poop
mwup
mmo
oocw
m—m

000-

~5mnN-

Fons

Popp—
oo—Nw
Nonnr

ovm
emu
who
cum
an5
mam

mph-

onwm-

«0.
com-

on
m~o
0&0
NF»
00—
o—
o—
pmp
me

o«5
ce—
005

—
cow-

o-~-

es
Mom
n~o
mmh
amp
or
or
Pm"
—mp

50
—o

o~P

NMN-

hump-

pcp-
onp-

-~
use
wnn
pno
mmp
op

on

Fmp
«mp

93!

mm—

5
non.

Poop-

50.
com-

amp
new
nnm
one
or.
or

o—

Pmp
Fmp

«o
«0

OFF

DON-

onnw-

cop-

co—
o—m
0mm
cow
055
05

0—

pm—
«m5

on
on

or—

0mm-

 

 

50.5.5

 

€05

 

aa2U5

Essa:

 

Ne45

 

 

.gumnea5

 

9:3,:

 

=mu<25

 

 

.2335 95 £23.:— - = .232
.3 ”.33...

 

99

 

owmn5
acum—
ocaN5
acw~5

cow
onN
anN
nNe
5n5

nhuwmo-

v5on-

Nnm
~n~

~m~

o~5
own
ann
cov

onh-

v5cn-

anN-

e5hv-

55a5
55a5
55o5
5505

:05
0mm
on~
o¢v

corn-

oase-

can
coo
oao
coo

anN
anN
nub

NchN-

mono-

can
oac
can
coo
vow
onN
onN
ans

Ncha-

N5wnN-

ash»
”hum
ash»
wh§w

n~5n-

Vboh
cues
whoh
vuos

om5

«MN
5cm

amnfl.

when
when
@605
VFOB
°~5
cmn
anN
5cm

ammu-

N9905-

Vbeh
vheb
vuou
vhoh

0N5

can
5o»

haun-

emce-

e50
050
v5e
v50
was

ON5
van
onw
nun

mmON-

onco-

v50
v50
v56
v5o
0N5
can
onu
man

amou-

owoo-

v50
#50
v50
v50

¢~5
«mm
on~
nan

nonm-

oven-

«N5-
vn5-
«N5-
QN5-

and
own

aneu-

VN5-
VN5-
¢N5-
VN5-

onu

anN
nNh

anew-

5nVnm-

5onn5
5onn5
5onn5
5enn5

5nnn

«MN

oran-

2: so...
mzmu
:9.
38m
3%.."
9:3.
£22:
80
2095»
3.5.
m5
960
:22.
m2:
mz<8
«<95

mcLuymu
mamymm
mwa‘mm
.mNa<<§n
UZ—Sr
025n>
UZ5~>
0255>
mawnumn5
n419>nza

nawyAMm
mafidn:5¢‘

 

 

 

5n=m¢

 

 

 

5aq9ux

 

naaxux

 

«aq3u1

 

5:2ynu

 

nczynm

 

angynz

 

595n—

 

9:.—

 

nu>r5

 

5enu¢

 

annum

 

«enuu

 

 

58a? 8.85 - magi—m ”55 £80 33.82:: 33th - 3:5 .082
m.» ”.339

 

100

 

25n-

«ev-
Even

8cm
855m

GQN
«am
can
can
52

man-

25n-

N31

can
can
8cm
man
can
can
3 5

ca"
mac-

«a—
5.5m
«mm

35

ans-

an
an
nmn
53

5.3”
amb-

can

«m5
v9.
am
can
3—

:5-

Sen

2535
an
an
mu»
5m—

can
3%-

h<m=>a A

8:.
38m
358

3:93
80
28.58
«5...:
:5
«<8
:28
m>qm
5.72%
ms...“—
93555
$55
9?...
szE
H.553):—
«no;
«mo?
52>
«no;
9:93
3:52
52.—Co
33.53
5:03.
:32.
9322.
3:2
5.4-2%.
:55:
:55...
9372..
a255—
zap—.5.

 

 

QUSNE

 

..50 RE:

 

“Um—$5

 

.50 55.53

 

hays."—

 

h@ 50”—

 

 

55:3 8:38:— 83: - Eébteu was 2830 Eat..— . 3:: .332

0.9 and?

 

101

 

hm.

mm.

coon-

Nov-

53
53.
53
53.

won
m5m
o5m
omm
5n5

now-

coon-

Nov-

53.
53
53
5:
2m
an
own
5m5

mam
now-

8mm-

moo 5
So 5
5585
Moo 5

35
gm
mmn
Nam
5m5

unh-

comm-

m2:
m2:
85:
meg
Rm
awn
man
5.». 5

35
m2.-

one
09
one
one

Nn5
3v
«5n
can
5m5

one.

one
amv
amv
amv
3v
a 5 m

505

um 5
2b-

Pg
M50555
QANmm
Q5 5 mm

So
2083
85
m5
960
.228
m>-E
9,28
33
3.2

San
m2<mm

02555
02559
025N>
025 5%
@385
2502

Emma
£50355.

@522
gé
3%

gem-.5
m<-52<n

255-5.:

 

 

moan

 

Raga

"50ng

 

 

“SENS

"50953

 

 

"505553

 

"508%

 

.5.-V5055

 

 

58.5.? 8.85 - Egbtfiw «£5 H.555 333.5 - 3:: 332

We Ham-Sm.

 

102

 

mm.-

mm.-

ma.-

2.-

mu.-

3.-

3.-

mm.-

mm.-

mm.-

mm.-

mm.-

m<50mfl
:02

Emma
52505.5(
g5.
9.152555.
:52
332%
acid—<52
gamu—
325.
DEC

2555

 

 

5 .535v

 

5n>o=

 

2.8

 

:33

 

Ema

 

23..

 

2.5.5

 

25.

 

2::

 

2.2.

 

5.5.502

 

255

 

 

8:52 9.3: .33

a.» mam—<5.

 

103

 

mm.

mm.

mm.

mm.

mm.

mm.

mm.

mm.

mm.

mm.

mm.

mm.

93qu
302

gnaw
Q503<
gé
3.2
gawk
D7215

Emu

 

 

Guam-.50

 

>02mh50

 

BOOM-.50

 

mum-"5.50

 

USED

 

ADE-.50

 

SO

 

><S5mn50

 

MAE-“EC

 

MESH-5.50

 

QED

 

EC

 

 

83:53 2% 8.3

a6 Ham—<5.

 

104

 

ow.-

om.

om.

om.

om.

NH.

ov.

mm.

mm.

 

hing

 

 

ngm

NZOM—m

 

Nngm

 

mn<Um

 

 

m<n¥0m

 

3520.5

 

#295

m>-5\.5m

 

 

gown

 

(wanw

 

gm

 

 

Ema

mz<unw

 

 

N23

N353

 

 

 

3:33 26 no.6
2.» 3:5.

 

105

 

fl

nfl

n”

n”

nN

mijzz
80
ZOUKDm
02$%
023%
023%
022%
ZdDPMM

 

 

 

88

 

vm<m%n

 

mm<m%a

NM<M%Q

 

 

5m<m%fl

 

 

85553594 8.3550

55.9 Hum—<5.

 

106

6. 2.3.3. Accounting Rows

Accounting rows are introduced for all crops to allow the model to transfer output
from the production to the sales or consumption activities. These rows allow modeling
production and sales separately. This approach facilitates sensitivity analysis. The accounting

rows appear in Tables 6.3 through 6.7 and 6.10; they are:

Dimba Maize DMAIZE
Beans BEANS
Rainfed peas RPEAS
Groundnuts GNUTS
Pegion Peas PPEAS
Sugarcane SCAN E
Pumpkin Leaves PLVS
Tomato TOMATO
Chinese Cabbage CCAB
Cabbage CAB
Rainfed Maize RMAIZE
Tomatoes Seasonl&3 TOM1&3
Onions Season 2 RON2
Wheat WHEAT
Rice RICE

6. 2.3.4 Deviation and Chance Constraint Rows
Income or maize yield constraint rows account for how enterprise risk enters the

objective function. These are:

First year incomes or maize yield YIINC or YlMZYD
Second year incomes or maize yield YZINC or YZMZYD
Third year incomes or maize yield YBINC or Y3MZYD
Fourth year incomes or maize yield Y4INC or Y4MZYD

Chance constraints rows are used to make sure the solution meets the probability limit. They
are the deviations constraint [Q(t)], and the sufficiency condition (SUFCON). The Q(t) sums
UP the value of the yearly deviations from a selected target weighted by the probability of the
occurrence of that year’s income or harvests. In this model the probabilities have been put at
0 - 25 for each of the four years. The sufficiency condition enforces the requirement that the
income achieved is not below the goal in more than the specified proportion of cropping

Seasons.

frc

des
cro

C01

 

 

 

107

6.3 Summary

This chapter describes the formulation of the model used to determine the optimal
mixes of activities. It begins with the description of the compilation of enterprise budgets
from which were derived many of the co-efficients that are needed for the programming
matrices. For crops that are currently grown, the survey data are used. For intensive and
irrigated enterprises sources of data are various government documents. Section 6.2
described the various components of the programming matrices. The activities included are
crop production and sale activities, labor hiring and sale activities, and deviation counters.

Constraints are family labor, land, accounting rows and chance and deviation rows.

Sll
0?
ine
rec.

van

 

 

 

CHAPTER 7: MODEL I - CURRENT CROPPING PRACTICES

In this chapter the results of Model I using income and maize production goals are
presented and discussed, including model validation. Model I deals with optimization of the
representative farm’s income with current crop packages. It is important to understand
whether farmers could be optimally allocating their resources given their present technology
and why they might not be doing so if that is the case.

The method used to derive the results presented here is described in detail in sections
3.2 (g) and 3.3; its formulation is documented in section 6.2. It is a Target MOTAD model
due to Tauer (1983), which uses T elser’s safety first probablistic criterion with Atwood’s
(1985) lower partial moments (LPM) to model yield or income risk in growing crops. To
recapitulate, the algorithm selects the activity mix which optimizes expected income while
simultaneously enforcing probablistic constraints upon the deviations of income from this
optimal expected income with the least constraining linear lower partial moment stochastic
inequality. The constraint is t - L*Q(t) > g, where t is target level of choice variable, L* is
reciprocal of the probability limit, Q(t) is the sum of deviations and g is the goal level of the
variable.

The modeler can parametrically change the probability level below which failure to
achieve goal is undesirable or change the goal itself. In this analysis exercise this probability
pertains to the proportion of cropping seasons above which it is unacceptable to fail to achieve
the specified food security goal.

With respect to the goal, the approach in this study is two pronged. There are two
main ways of looking at the solutions to the maize insecurity situation. A farmer can try to
formulate a plan that allows her to produce all maize, or she can produce some maize and
also cash crops to earn income to make up for the deficit. Each of the three models is
presented in two variations: the goal in the sufficiency condition is specified in kilograms of

maize harvested and also income (gross margins) from all enterprises. In case of maize

108

M

USE

COn

on It

 

 

109
production, the goal is 1,000 kilograms of maize: 200 kg per person per year for a
household of five. These estimates are based on government sources [Malawi Government
1993(b)]. Minimum income requirements are calculated to be MK2,500 per household per

annum. The following budget was used to arrive at the datum:

Item Costhk)
Maize (200 Kg @ MKO.60/kg) 120.00
Salt 20.00
Cooking Oil (4 liters @MK10.00/literl 40.00
Sugar (10 Kg @ MK2.00/kg)' 20.00
Protein Source (@ MK.50/day) 180.00
Vegetables (@ MK0.05/day) 20.00
Transport/Sundries/Hospital Fees 100.00

Total 500.00

Assuming an average household size of five, the minimum income requirement is
MK2500.

Some government/donor documents [Malawi Government 1993(b), World Bank 1990]
use a poverty datum of US$40.00 which was MK200 during the time this study was being
conducted. This is very low and has neither empirical basis nor description of its derivation.
Each specification of the two goal specifications is analyzed for all three farm sizes. The
target, t, is endogenously selected by the algorithm but it should be greater than or equal to g.

Sensitivity analysis is done to understand the impact of selected changes in variables
on the optimal solutions. It is done using non-marginal changes in values of these variables:
values outside optimality or feasibility ranges for optimal solutions. The variables that are
parameterized are prices of output, cost of hired labor, yields, and cost of irrigation. When
planning investments, the relevant price for valuing output is the one to obtain when the
planned production enters the markets. This price may not be discovered in advance; thus,
sensitivity analysis evaluates the impact of conjectured price changes. Even with irrigation,

yields can vary for the same season in different years and between seasons due to water

sour
diff:
the I
Dec
in k
abse
redu

that

solu
of ft

inco

'9
v-

 

 

 

110

source fluctuations, management of crops and natural disasters. There are potentially five

different irrigation methods with three cost levels. Sensitivity analysis is intended to capture

the difference in costs of these methods. Cost of labor varies by month. In the months of

December to January many members of small farm households do "ganyu", work for payment

in kind, on larger farms. The wages are believed to be below the minimum wage. In the

absence of empirical data on the level of these wages, sensitivity analysis is done for 50%

reduction in the minimum wage. It is done for Model III(a) only, because it is in this model

that there is significantly higher labor hiring.

The results presented and analyzed include I) the level of activities in the optimal
solutions, 2) shadow prices of resources and constraints, 3) stability of solutions and the cost
of forcing excluded activities into the solution, 4) contributions of various resources to total
income and 5) highest feasible maize production goals.

1) The optimal solutions are of two kinds: non-risk constrained and safety first
solutions, where applicable. The results that are discussed extensively are those at the
probability limit of 0.25 (L*= 4). These results can be achieved by representative
farms in at least 75% of the cropping seasons. However, also presented are results
for the 0.33 and 0.40 probability limits for small farms. This is done to illustrate the
general nature of solutions and provide supporting evidence for conclusions reached.

2) Shadow prices measure the rate of change in the value of the objective function in
response to a small change in the use of a resource or level of constraint. Alterna-
tively they can be seen as the maximum price to pay to get one more unit of a
resource or otherwise relax a constraint by one unit. This is true if the resources are
not sunk costs. Shadow prices also indicate the productivity or scarcity of resources
available to the farm. The more constraining the resource, the higher the shadow
price.

3) Two concepts of stability will be used. The first is range of optimality (Anderson et

al 1991). This refers to the range of values of the objective function co-efficients

4)

5)

111

over which the activities in the solution function remain optimal. The second, range
of feasibility, indicates the range of the right hand side values of resources/constraints
over which the dual value will remain the same. Cost of forcing in excluded activities
in the solution reflects the competitiveness of the activities. Competitiveness is
defined as the cost to a farmer of forcing in an activity excluded from the optimal
solution. The higher the value the less competitive the activity.

For all solutions there is a discussion of the contribution of various resources to total
income viz, dimba land, rainfed land, irrigated and off-farm income earning activities.
These data provide a way of assessing the relative importance of various resources in
to achieving food security goals.

For each model there is also a presentation of the highest feasible maize production
goals and the corresponding optimal maize production levels. The maximum feasible
goals show the level of production that a farmer can hope to exceed in 75 % of the
cropping seasons. They indicate the level of confidence that farmers can have about

their potential for maize self-sufficiency.

7 .1 Base Solutions

7.1.1 Maize Production Approach and the Non-risk Constrained Solutions

7.1.1.1 Activities in the Solutions

The general pattern of solutions is shown in Table 7.1. The non-risk constrained

solution and the maize production approach give the same results because the same activity

has the highest gross returns; maize/beans/pigeon peas also has highest maize yield. The

level of goals are selected to represent the ends and middle1 of the feasible but binding region

of the sufficiency condition. At all probability levels, the value of the objective function

 

' In Model I the results do not show a lot of variation in level of optimal crops so that

only the ends of the spectrum are presented. Table 7.1 can be compared to Table 8.1 for
Model II.

3.:QO mama—R32 San— mutsm 528m

eggucoo. $5653 Euwctu
82: 2: c8 323 :9. fl :2on 037.80 05 .3526: ..aow Engme 2.55:6 a was 3:53qu some 20:3 _.w 292. 523 @9335”. up
:8 033 m3... .cnmémm 823 mm 5&2 mficfin 05 .36 .wd .328: SE»: “a 85 .04 .232 523305 025 2.: E 853006 8.88 < R

112

 

 

 

 

..... 2 .o mvd --t E. 36 $2 2% owed
..... wfl .o mud ---- E. 36 :2 «Ohm ovd
ll. 2 .o 36 :1 i- mvd mm? ohm Sod
afi .o med t: t: 3.0 32 .Nwm mmd

Ed Q .o 93 t: t: 3.0 amfl Eh
Sod ammo med t: --t 36 whfl 5% mad
t: 26 36 ed od $6 8.2 o =<
0:20 0E6 *4:
0885 w :85
288. venom admin—M 29:32 25552 "Emma: :32 ~80 .905

38385 83323.. 35
«8.8m :55

23 23 23 .25 £85 £3.65 2 sense: 3:8 533 ”.8032 as
senate 053305 ”.5209 was: 82.28 .35 beam - _ 252
E 0:3

fl. Ni p0~fi-nt~t .ri .-qvflwtnv l h ~.v‘b.-<

 

113

4mm 4mm
we we
omm cm~
8m 8m
3. :
a. we
own ova
m3 moo
93 26
mi
a was
me: o
ER 23

Hwy—Em ZOUMZ
033

8 cm
OR OR
EN EN

mm Vm

we 8
Ohm ohm
8m 8m
3w _ b:
No“ ma
mm _ mm H
ovm 9%

3 ca

2 .o C .o
9.6 9.6
Nwd ll
36 co;
moa o
and; 5mm;

Exam zoomz
5?:

o2 $2
3N vwm
0mm emn
52 02
b: m3
SN hwm
Nam 8m
ovm Sew
ll 2N
mom mom
ova ova
a2 52
5.9 l:
N— .o 3 .o
36 5.6
med mvd
2% c
own; own;

Ewan—m .ZOUMZ
=~Em

.: . 5:50
3588R—
“38302
c3300
caesium
.mzw:<
33.
0:3.
>52
=a<
5.82
Sign
Egan
Amazons—8.8% 3.0— gumbo

2:5...
ogawsm
as“. BEB—
"Emma:
29%:
mean:
3:08:32 8:885 86

0885 :30
0.200885 5»:
€62

 

:8: $538.. as 2: 2 cacao: 35m 333 $832 as
8:25 32525 £33 a? 283% EE 5% - _ .232
2 use.

sous—om 5 8238.4 no 63‘—

114

3150“ wwmitmgfi San— ~Au>hflm ”nouflom

82.2% EE SE n Exam 3

mecca—om £80394 cocoa—coca 332 8 258 2a muons—om 3:35:00 meA—oz 2: "20052 \a

 

ma
3N
cmN

wmm
3

thm

mam
own
.5
03
«Fun

ma
2%
0mm
com
2 _
03

3

thm
mam
a5

SN
chvm

mam“
>2

or
m:
cm:

mg
on
2.

3
mm
2m

unaccoun—
Sna— “caesium
:53 5:92
.892 32.
>22
E3.
baa—an
350:...835 moE>uu< 8%..— 3::

unauuawsm

80388.

30A 30:5—

82.2580

anon noowE

88m

382 BE
Agassi 82.3. E...“

3.5 E 23

CC

Cll

Cal

Ma

115

when the sufficiency condition is non-binding (due to a zero level goal) is the same,
MK1,380. In this particular case it also happens that the maximum feasible income goal
(MK576) is the same at all the selected probability levels. Due to the narrow band of the
feasible but binding region, the solutions do not reflect much difference in crop combinations
and values of the objective function. This is due to the tightness of the risk situation as
represented by the sufficiency condition.
The results of the base scenario at the 0.25 probability limit are shown in Table 7.2.
The configuration of yearly yields and output prices leads to the same solution for the safety
first maize production approach and the non-risk constrained solution. This is due to the
mean-minimum dominance of the fertilized maize/beans/pigeon peas enterprise among rainfed
crops. It has the highest minimum yields and the highest mean income. Since the maize
production approach optimizes expected income subject to a safety first constraint on yield
deviation, maize/beans/pigeon peas is selected. Sugarcane is optimal for dimba land and a
relay crop of field peas. The level of activities in solution-production and sellinguvaries with
available resources for the three farm sizes. (It is assumed that the farmers will sell what they
produce over consumption requirements.)

The maximum mean returns are MK1,380, MK1,657 and MK2,529 respectively for
the small, medium, and large farm sizes. The representative small and medium farms meet
only 52% and 66% of their subsistence income requirement of MK2,500. These results
corroborate the findings of other research studies on levels of malnutrition and poor health as
cited earlier.

Small and medium farms have surplus labor which they can sell or use in income
earning activities (IEAs), every month. Large farms have surplus family labor in February,
Mar Ch, May, July, October and November. However, they need to hire labor in January,

April , June, August, September and December.

\l

C0

the

PIC

 

1 16
7.1.1.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises
The shadow prices of rainfed land and dimba and relay use of rainfed land are
NIK196 and MK3,338 and MK41 respectively. Thus, under the base scenario enterprise price
and yield permutation, dimba land can generate MK3,338. This dual value of land is feasible
over a range of 0 to 0.46 ha, 0.17 to 0.23 ha, and 0.32 to 0.55 ha with respect to the three
farm sizes. Sugarcane is optimal until prices drop below MKO.28/cane. Sensitivity analysis
is thus done at MKO.25/cane.
On dimba land there are no crops that compete well with those in the optimal solution
on small and medium farms. Introducing any dimba crop not already in the solution can
completely wipe out the gross returns. Fertilized maize/beans/groundnuts is the most

competitive on rainfed land at zero cost of being forced into solution.

7. 1.1.3 Contribution of Dimba Crops to Income
Dimba crops contribute 31%, 38%, and 47% to the optimal income of small, medium

and large households respectively. Rainfed land is most important to large farms,
contributing 41% of their income (See Table 7.3). This result is important. From the small
portions of dimba land, the small representative farms can generate more income than from
their rainfed land. For the representative medium farm a 0.17 hectare plot of dimba land can
provides the same income as 1 hectare of rainfed piece of land. These results have two
iIIIplications. First, dimba land is very important to providing livelihoods especially for small
fal'rns when compared to the other farm sizes. Second, small increases in availability of
ditnba land can have significant positive impact on household incomes. A conclusion can be
made that when favorable market conditions exist, upgrading small portions of rainfed land to

irrigated status has the potential to significantly improve the incomes of households, and their

food security status.
Farmers can generate a substantial portion of their income from non-farm IEAs:

49% . 31% and 12% for small, medium and large farms. Small farms can thus derive nearly

l 17

half of their total income from use of agricultural slack labor in either IEAs or employment.
This study did not focus on off-farm IEAs. A good specification of the off-farm IEAs is
important to draw specific conclusions. However, the finding is in line with the descriptive
results as presented in Chapter 5—that small farms tend to be engaged more in off-farm IEAs
than large farm operators. Studies conducted in other countries also have similar findings:
Flanders (1977), Priebprom (1982) in Thailand, and Reddy and Finders (1988) in USA.

This finding indicates that if farmers with very small holdings are not able to practice
intensive agriculture, then government needs to focus attention on off-farm sources of income

either to make them more remunerative or easier to establish.

Table 7 .3
Model I - Base Scenario
Contribution of Dimba Land, Rainfed Land and Off-farm IEAs to Income

 

 

Size of Farm
Small Medium Large
Rainfed Land 20% 31% 41%
Dimba Land 31% 38% 47 %
Off-Farm IEAS 49% 31% 12%
Total Income (MK) 1380 1657 2529

‘

Source: Survey Data l993/LP88 Results

118

7.1. I . 4 Maximum Feasible Maize Production

While the non-risk constrained solution does not change with probability limits, the
level of achievable maize goals do change. These are indicated in Table 7.4. At the 0.25
probability limit, the feasible maize production goals are 300 kg and 708 kg, for small and
medium farms. The corresponding expected production levels are 668 kg and 1,480 kg.
Thus, small farms can be sure that the expected maize production of 668 kg will be more than
300 kg in 75% of the cropping seasons. In other words there is certainty that small farms can
provide adequate maize for 1.5 persons in three quarters of the cropping season. Large farms
can meet the safety first goal of 1,000 kg without being constrained. When the proportion of

cropping seasons permissable to fail to meet the goal is raised to 40% , the medium farms still

fail to reach a mean of 1,000 kg.

Table 7.4
Model I - Maximum Feasible Maize Production (kg) Goals
by Allowable Probability of Failure

 

 

 

Size of Farm
Probability Small Medium Large
Limit
0 - 25 300 708 1 .663
0-33 331 780 1,834
0-40 405 954 2,401

 

 

I“Ote: The minimum maize requirements are estimated at 1,000 Kilograms per household

119

7.1.2 Safety First Solutions

In this section the results of the safety first solutions at the 0.25 probability limit are

discussed. These were shown in Table 7.2.

7.1.2.] Activities in the Solutions

The crop mixtures vary by farm size. Small farms have fertilized maize/beans/
pigeon peas on rainfed land and sugarcane and a small amount of tomato on dimba gardens.
Medium farms have fertilized maize/beans/groundnuts on rainfed land and a small portion of
fertilized maize/beans/pigeon peas. Large farms have fertilized maize/beans/groundnuts.
Sugarcane is the sole crop on dimba land for both medium and large farms. The differences
in solutions are due to the resource endowments of the three farm sizes. When safety first
goals are specified in income and land is very limiting, activities with very high means returns
remain in solution, i.e. maize/beans/pigeon peas and tomatoes for small farms. On large and
medium farms, maize/beans/groundnuts dominates.

The highest feasible income goals are MK576, MK903 and MK1,863 at the 0.25
probability limit, for the three farm sizes. They correspond to incomes of MK1,359,
“1,638 and MK2,471 with respect to farm size. None of the three representative farm
Sizes can achieve incomes that are higher than the minimum income requirements of MK2,500
in 75 % of the cropping seasons.

Only the large farms need to hire labor to support their crop activities. This is in the
months of January, April, August, September, and December. The medium and small farms
can sell family labor in every month because the sizes of their gardens are too small to absorb
it all -

A slightly different version of Model I was analyzed with hybrid maize/beans/pigeon
peas - This model corresponds to the past extension practice of offering credit for only hybrid

maize. When farmers are risk averse, this enterprise does not appear in optimal solution.

0f

0n

120

7. 1.2.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises

From the non-risk to the risk constrained solutions, the dual values of land change,
rising from MK196 to MK524 for rainfed land; MK3,338 to MK4,943 for dimba land; and
from MK41 to MK115 for relay use of rainfed land (MK82 for small farms) (see Table 7.5).
The dual values of land are very high compared to rental or sale rates, even though these are
not well established empirically. The dual values are derived as the net return to the optimal
activity grown on the land. They, thus, represent the income the farmer loses when not using
the land, or income to be had from using one more unit of land.

The differences in the dual values of land in the non-risk constrained solution and the
risk constrained solution is the safety first premium of land. It measures the contribution of
one more unit of land to manage risk as measured by the yield/income variability and
perceived by the farmer. The latter is reflected in the sufficiency condition (goal and

probability level).

The range of optimality is narrower under the risk constrained solutions than the non-
risk constrained solutions or maize production approach. Thus, risk averse farmers may
watch price changes more carefully because they will impact their safety first goals. The dual

of the sufficiency condition of 0.54 shows that the expected value of the objective function
can increase by MKO.54 if the goal is relaxed by MKLOO. (The value of the dual of the
sufficiency condition changes with probability levels, the value being lower at higher
Probability levels. Thus, at higher probability limits (l/L), there is less trade off between
i“Conic safety first goals and returns.)

The most competitive alternative crops vary by farm size. For small farms these are
cabbffige at zero cost per hectare on dimba land. On rainfed land there is no competition from
any crop. On medium and large farms, the dimba crop of sugarcane has no competition. In
case of rainfed land, non-fertilized maize/beans/pigeon peas could be forced in solution at cost
of MK23 per hectare on both medium and large farms, and fertilized maize/beans/pigeon peas

on 1al‘ge farms at zero cost per hectare.

121

7.1.2.3 Contribution of Dimba Crops to Income

The configuration of sources of income from dimba, rainfed and labor are similar to

the non-risk constrained solution.

7.1.2.4 Maximum Feasible Maize Production

The maize yields obtainable when income safety first goals are binding are 628 kg,
1,251 kg and 2,779 kg for small, medium and large farm sizes. They are lower than in the
case of no risk aversion or maize approach because the criteria for selecting optimal activities

is now based primarily on income variability rather than mean returns. The activities with

less income variability yield less maize per hectare.

7 .2 Sensitivity Analysis

Sensitivity analysis is done for three cases. The first, Scenario A, allows the price of
the dominant dimba crop in the base solution, sugarcane, to drop by 50% from MKO.50 to
MKO.25. Scenario B assesses the impact of mean yields of the tomato enterprise at the price
of MK1.00 on expected returns and optimal crop mixtures. (The base scenario has median
yields of tomato). Scenario C reduces the price of tomatoes in scenario B by 50%. This last
scenario corresponds to the situation where high yields of tomatoes can glut the market or to
account for high perishability of tomatoes. The results of sensitivity analysis for small farms

only are shown in Table 7.6. (Results for medium and large farms are similar with respect to

Combinations.)

122

Table 7.5

Model I — Shadow Prices of Resource and Constraints

 

Resource/Constraint
Small Farms
Land (MK/Ha)
Rainfed Land
Relay Rainfed Land
Dimba Land
Labor All months

(MK/Person-hour)

Sufficiency Condition

SFIRST'I
Medium/Large Farms
524 241
1 15 83
5,526 4,943
0.25 02.5
-0.71 -0.54

NRCON"
All Farms

195
41
3,338
0.25

 

a! Safety First Solutions

b/ Non-risk Constrained Solutions

Source: Survey Data l993/LP88 Results

:. “quIIIovj

.CK. 07.3%

123

238M wwmdmoa— 2mm hut—am ”bosom

Sosa—om 322380 22282 ”ZOUMZ ”82.28 2ch .028. ”Hwy—Em E

350922 22:2 co out: 23 .2222 22:2 :82 $528622 383?; no out: “U 228on 2
350.3%,“ 22:2 co 8:: 23 .2203 22:2 :88 .oSWQmNdME 883?; «o 025 ”m 2.3:on B

3525:)“ 22:2 co out: :5 £an 22:2 5:2: .ofioRNdMZ 0:832; «o 025 ”< 23:on B
wctoogvaz 22:2 .8 out: 25 £22» 22:2 22:2: 6.828622 05833. co 02.5 ”2.85% been \a

 

 

 

 

cod 8.9 8: 2:- o $2. 2083
2:2 8.3. e: G e: 3 23 3e:
8.92.2 8.23 5.2 :5 a}. N2; 23 SEE
Bead 8.3: wen me m3“ 3: 23.2
33> :25
Se m3 m3 fie:
mg m; 225:.
w _ .o 2 .o E. .3 E. i- ogcamsm
2 E. 220
3.: 3o 93 3o me: 98 38:
e e ”.522
3.: 38 E. 2222
as $3 nee nee "flame:
8_._>:o< :80
we? 23 no} 2} 23 23 282: 882m
52% 28:2 .525 2822 Exam 28:2
.0 235% am 235% .2 258m
/

Ea: .Eem - 3382 2235 228% - _ 282
es 2.3

124
7.2.1 The Non-risk Constrained Solutions

7. 2. 1. 1 Activities in Solutions

In all the three scenarios, fertilized maize/beans/pigeon peas is dominant on rainfed
land in the non-risk constrained solutions. On dimba land, maize is dominant in Scenario A,
tomatoes in Scenario B and sugarcane in Scenario C. The maximum mean returns for these

three cases are MK1,158, MK1,482 and MK1,380. The base solution has an expected

income of MK1,380.

7. 2. 1. 2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises

The shadow price of dimba land has gone down from MK3,338 in base scenario to
MK1,632 and MK1,421 in Scenarios A and B respectively. This is due to the low profit-
ability of dimba maize and tomato at the prices and yields in these scenarios. Values of the
sufficiency condition are higher for Scenario A and B, reflecting that its relaxing it by one
unit would add more to the expected value of the objective function than in the base or
Scenario C.

Under Scenario A, sugarcane competes with dimba maize at zero cost per hectare on
small farms; tomatoes at cost of MK37 per hectare on medium farms; and sugarcane and

tomato both at zero cost per hectare on large farms. These results show that the kinds of crop
mixes grown by farmers can be varied among dimba maize, tomato and sugarcane when the
price of sugarcane is low. When it is high, as in the base solution, sugarcane is dominant.
On rainfed land, maize/beans/groundnuts can be exchanged with fertilized maize/beans/pigeon
peas at a cost of MK24/ha. The optimal crop mixtures are: maize/beans/pigeon peas, rainfed
peas for rainfed land in all scenarios; on dimba land maize for Scenario A; tomato for
Scenario B or sugarcane for Scenario C. Thus, fertilized maize/beans/pigeon peas is
dominant under all scenarios with no risk aversion. But each scenario has a unique dimba

crop.

125

7. 2. 1.3 Contribution of Dimba Crops to Income

The proportional contribution of dimba crops to income under the three scenarios is
shown in Table 7.7. The base scenario has a sugarcane price of MKO.50 and median tomato
yields at MK1.00. Scenario A has a MKO.25 sugarcane price and median yields of tomato at
a price of MK1.00. The contribution of dimba land falls from 35% in the base solution to
30% in Scenario A. Rainfed land contributes only 19% and off-farm IEAs 51% in Scenario
A.

In Scenario B, the price of sugarcane is at MKO.25 and mean tomato yield at
MK1.00. Tomatoes are in the optimal solution. The contribution of dimba land is 48% and
that of rainfed land 15 %. It is thus demonstrated that the contribution of dimba land to
income is dependent on the price of what is grown on it. However, under each scenario the

contribution of dimba crops to income is considerable.

Table 7.7
Model I - Contribution of Dimba, and Rainfed Land Crops
and Off-Farm IEAs to Total Income Under Three Scenarios
Small Farms :

 

Base Solution Scenario A Scenario B Scenario C

 

Expected Income (MK) 138 1 158 1482 1380
Rainfed Land 16% 19% 15% 16%
Dimba Land 35 % 30 % 48 % 35 %
Off-farm IEAs 49% 51% 36% 49%

 

a/ The figures are for the NRCON Solutions. The difference in proportions between the
NRCON and Safety First solutions is mostly I to 2 percent.

Source: Survey Data 1993

126
7 .2.2 Safety First Solutions
7. 2. 2.] Activities in Solutions at the 0. 25 Probability Limit

Scenarios A and B have non-fertilized maize/beans/pigeon peas on rainfed land.
Scenario A has maize and Scenario B has tomato on dimba gardens. Maize/beans/pigeon peas
is sill optimal on rainfed land in Scenario C and sugarcane dominates in dimba land. There is
a small amount of cabbage and dimba maize. These crop combinations generate incomes of

MK1,118, MK1,455 and MK1,348 for the small farms with respect to the scenarios.

7. 2. 2.2 Shadow Prices, Stability of Solutions and Competitiveness of

Enterprises

The shadow prices of rainfed land are MK348 for Scenarios A and B, and MK2,296
in Scenario C. The much higher dual value in Scenario C is due to fertilized maize. For
dimba land, these are MK4,481, MK10,151 and MK18,330. The difference is due to the
profit that optimal crops in each scenario can generate. In Scenario A, tomatoes, sugarcane
and cabbage can substitute each other depending on which crop is in the optimal solution. In
Scenario B only cabbage competes with tomatoes at zero cost per hectare. Under Scenario C

there in no competition to the sugarcane optimal solution on dimba land.

7. 2.2.3 The Contribution of Dimba Crops to Income

The pattern here is the same as that for non-risk constrained solutions. The level of
contribution depends on the prices and yields assumed. Scenario B has highest contribution

from dimba crops (48 %).

7.3 Lessons from Model I

The foremost lesson from Model I is the contrast between the results of the maize

Production and income approaches. The first contrast is the level of incomes. Those of the

127

income approach are lower when the sufficiency condition is binding. The second difference
is the optimal crop mixes. The non-risk constrained solutions have fertilized
maize/beans/pigeon peas, whereas when safety first goals are constraining the non-fertilized
maize/beans/pegion peas is optimal.

With respect to cropping patterns, it was shown that when farmers are risk averse,
stable and higher incomes from dimba crops can make fertilizing maize optimal. In the base
scenario and also in Scenario C, the higher value of sugarcane, a very yield stable crop,
enables fertilized maize to be optimal. However, under this model only local maize is
optimal, not hybrid varieties. This may explain the disappointing past experience of non-
adoption of hybrid varieties. Credit was made available only for hybrid maize and other
rainfed crops. This did not solve the basic problem of income risk due to yield risk because
all crops for which credit was available are subject to yield risk and have very low gross
margins in bad years. Efforts to increase incomes from dimba cultivation can have the
desired impact of making farmers adopt fertilizer on maize. Therefore, credit and extension
programs need not be maize centered to encourage fertilizing for maize.

Under current cropping practices, when a) farmers are very risk averse, or b) they
base their safety goals on income (i.e. perceive fertilized maize as a cash crop), and have
unstable and low non-rainfed land sources of income, they should not fertilize rainfed maize.
The fact that a large proportion of farmers do fertilize their maize indicates that they have
other sources of income to manage the risk of fertilizing maize or else they perceive fertilized
maize as a food producing activity. Both of these factors are at work in this area. The
analysis has shown dimba land can provide higher incomes. This plus the proximity of the
large urban area of Limbe-Blantyre provides a reliable market for dimba crops for farmers
from this area. Wage employment may also be a factor in the stabilizing incomes.

The contribution of dimba crops to income varies from 35 % to 48 % depending on size
of farm. It also depends on the value of crops grown. Small sized farms could derive half of

their income from off-farm lGAs.

128

7 .4 Validating the Model

Model 1 represents the current situation of farmers. The results of the model indicate
that it performs reasonably well when compared to the descriptive results reported in Chapter
5. First, the activities in the solutions as well as the most competitive crops are dominant in
farmers’ cropping patterns. These are maize based crop mixtures and relay peas on rainfed
land, and with sugarcane, tomatoes, dimba maize, and cabbage in dimba gardens. The
solutions to the base scenario and sensitivity analysis show that the model farms adjust to
grow different crops depending on price relations. This implies that careful market studies
should be done to assess the market and effective prices for crops in order to ascertain the
optimal crop mixtures farmers should grow in different areas and at different times. It also
implies that some method of managing supply is needed given the thin vegetable markets and
related price volatility.

The labor hired depends on the cropping mix in the optimal solution. Growing
tomatoes or dimba maize calls for more hired labor than growing sugarcane. The LP results
prescribe hiring labor only for large farms. In the survey results, all farm sizes hire labor,
but the frequency is higher among larger farms. The difference between the LP and survey
results may be explained by the observation that more adults from small sized farms
households tend to work off-farm. In these households, hiring labor is used to substitute for
the working persons.

Another plausible explanation concerns the estimation of labor available. In this
analysis, it was assumed that the equivalent of two adults are available to work 5 hours a day
during rainy season and 6 hours in the dry season, 25 days in a month. These figures apply
to all farm sizes. Some survey results indicate that small farms have less labor. For
example, the 1984/85 annual survey of agriculture showed than the labor available varies
from 510 to 1,548 person hours with respect to farm size (Malawi Government 1985). What
is not clear in these reports is how the labor available was assessed. Assuming that only one

adult is available to work the whole year, for only 3 hours a day gives about 1,000 person

129

hours. It appears that the hours reported as available are hours actually worked. In the LP
results, small farms have so much agricultural slack labor that even if the amount available is
reduced to half, they would not need to hire labor. This suggests that small farms may be
using more than half of their available labor time on IEAs and, thus, need to hire labor to
meet the requirements of cropping activities. In this part of Blantyre ADD farms take up
seasonal wage labor in factories like those that process tobacco.

Optimal solutions always include intercropped enterprises on rainfed land. This is a
dominant practice in the area. Risk averse behavior favors diversification, and crops with
high minimum returns which, in many cases, have low average gross margins. These are
most likely to be enterprises that do not use fertilizer.

To sum up, except for the case of hiring labor, the model tracks very well what the
farmers are currently doing. It provides a rationale for the variety of cropping patterns that
are observed. The variety of cropping mixes can be explained by difference in the level of

safety first goals and crop yields and their prices.

CHAPTER 8: MODEL II - CURRENT AND INTENSIVE CROPPING PRACTICES

In this chapter the results of Model II are presented, discussed and compared to those
of Model 1. Model II supplements Model I with the opportunities for farmers to upgrade
their cropping practices from current to intensive status by using more chemicals, e.g. Copper
Oxychloride, D.D.T., Sevin and Ripcord and fertilizers such as 2012020, Calcium Ammonium
Nitrate, Double Ammonium Nitrate and Sulphate of Ammonium. They also use more labor
on intensive enterprises. This is done without irrigation. Results of Model 11 should throw
light on whether intensive cropping practices are optimal vis—a-vis current practices given that
they require more expenditures on inputs and use of more labor. They also provide a base
for measuring the worth of irrigation investments. Hybrid maize is included as an intensive

activity intercropped with beans and pigeon peas.

8. 1. 1 Non -Risk Constrained Solutions and the Maize Production Approach

8. 1.1.1 Activities in Solutions

The general pattern of solutions at the three probability levels is shown in Table 8.1
for small farms only. The non-risk constrained solution and the maize production approach
give the same results as in Model 1. Hybrid maize has the highest maize yields and the
highest gross returns on rainfed land.

It is clear that hybrid maize combinations are optimal only at low levels of risk
aversion (low safety goals or low probability limits). When risk aversion increases, intensive
maize/beans/groundnuts is optimal on rainfed land. Hybrid maize is not optimal because of
very low returns in years of drought. On dimba land, mixtures of the two crops, cabbage and
chinese cabbage, vary depending on the goal and probability levels. Cabbage enters the
solution in small amounts at high level of risk aversion because it has high yields in year 1

when chinese cabbage has low yields. Such crop diversification helps to manage risk.

130

131
The results pertaining to the 0.25 probability limit are in Table 8.2. When risk
aversion is low, the farmers meet maximum feasible income goals with intensive
maize/beans/pigeon peas and chinese cabbage. The maximum mean returns are MK2,261,

MK2,898 and MK5,019 for the small, medium and large farms respectively.

132

2222:: :28 b?
:2293 :2: o: co 52:23 2 35%...on «EH 3
as: 5:382: 2: a a new e52: 6:852 a

338: wee-=82 ea: 255 H852

“Em

 

 

mod mod med med -- 2: .N .32.;
Em :6 med med -- emmd owe;
-- 2 .o med cud 2 .o wmm.m own; ovd
:d 8.0 med med -- a: a .N .02.;
56 N: .o 2.6 3.6 -- m-.N can;
-- Q .o med Ed wmd 8~.~ own; mmd
:6 mod med med -- Q _ .N ewe.—
vod mod med med -- Sm.~ ova;
-- mfio med wad Ed mmmfi o2; mmd
-- 2 .o med 1.. med $~.N no =<
0:6 0:6 .5:
028:— » 5:5
20— :82 mace":— :Om~$= Emu—z: Buooexm :30 .905

 

moE>uio< :20

 

Em =25 - cad e5 mmd find 2:2..— 323305 2
25822 35m 53:..— m.eoo5< e5

8:25 22535 page... use: 22.28 .2: 23 - = .282

fiw 055.

133

5:32:00

 

:2 :e :2 :2 8:88:

:m: 8: 8: 2.: a: 3: .8882
:m: :2 8: 8: a: a: 8880
:2 :: 2: :2 8:838:
m: :2 .822
:m: :2 2: :2 2::
:8 :2 :2 :V 8e:
: :v .1 .22
m: :2 8 _::<
2. w: :2 :2 8: 8: 882
2.: :v: :v: :v: :v: 2:38.:
:v: :2 :e :2 :2 238:

$209522: 20— Scam-t0
mm: 2.: :.: :8:
..- mm: 5.: 2.: 8.: 2.: e8:
2.: 8.: e3 8.: 3.: 3.: e8: 82::
2.: i. 8.: 2.: 8:85:
2.: 8.: 1.. 3: 88:8:

3: 822.2 8882: :20

:8: : 5:: : :2: : 22: 288: :80

2:: 2:: :8: 8:: :2: 8:: 0:): 288: 8.82m

$.52": 28:2 5:2: 28:2 22:2: .zou:z 2:8,:

8:3 8:8: =8:

 

Eons—om :2 82323: mo :33

:5: £582: m:.: 2: 2
2:020: 35m 8304 2:833. :5
8:25 82388: page: 58 e888: .2: 8:8 - : 22:2:
:.: 28:

134

3:53— wwmdmma :25 mutsm ”8.50m

QEE om:— 8: 8:395 :mt : 6:85 8: :8: :8:
bzfigoa mud on: H: 03.334 .8 Row :5 .EEH .:o_2o>: :32 3 35:59.8 go: a mafia owe:— Sm cog—o: 35 bong 2: >1.

muons—om Hmém B

80:28 5:83: cases—USA 3:2 :8 m: 08:: on: 2: - Eons—om 358380 :mtéoz \:

 

ms
nu
mac
wmm
m:
2 n
how
Mb

mmwo
mam
Em
mmm
wawv

ma
3N

com

m:
om:
no

mnmm

mam

at
cum
:3“

avvm
_ w

53

mi

23

5mm
on:

$9:

52

ca
SN
Nam

Nag

598809
598033
“3&3:
b3.
83.
:2
:22.
5:5:
35:: cop—on: 8:353: 83::— 3:5

ow:nn:0

ow:nn:0 085:0

30m 335—

3:55.80

80:: coowE

88m

0:32 c853—
Aaflmofi: 83:3 was»:

8.288 N.» 03:.H

135
With this result, only the representative large farms may achieve returns that are higher than

the minimum income requirement of MK2,500 in 75% of cropping seasons.

8.]. 1.2 Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises

The shadow prices of land are MK384, MK9,463 and MK41 for rainfed land, dimba
and relay rainfed land respectively (Table 8.3). The shadow prices of land have shot up in
Model 11 by 97% for rainfed land and 183% for dimba land. The dual value for relay use of
rainfed land stays the same because the field peas crop does not have recommended intensive
practices. These values show a dramatic increase in the value of land if intensive cropping
practices are followed.

On rainfed land, intensive maize/beans/groundnuts competes with hybrid
maize/beans/pigeon peas at zero cost of forcing into solution, except on medium farms where
it would cost a meager MKl l/ha. This shows that even when farmers are not risk averse,
hybrid maize has serious competition from fertilized local maize based crop mixtures. This is
due to low on—farm yields of these hybrid maize varieties. On dimba land, tomato competes
with chinese cabbage on small farms and medium farms; and on large farms, cabbage and

tomato compete with chinese cabbage—all of these at zero cost per hectare.

136

Table 8.3
Model II - Shadow Prices of Resource and Constraints

 

Resource/Constraint SFIRST' NRCONb
Land (MK/Ha)
Rainfed Land 485.00 384.00
Relay Rainfed Land 88.00 41.00
Dimba Land 1,342.00 9,463.00
Labor All Months
(MK/Person-hour) 0.25 0.25
Sufficiency Condition -0.51 0

 

a/ SFIRST = Safety First Solutions
b/ NRCON = Non-risk Constrained Solution

Source: Farm Survey 1993/LP88 Results

8.1.1.3 Contribution of Dimba Crops to Income

In Model II, dimba crops contribute more to total income than in Model I. The
results are displayed in Table 8.4. First, income increases by 64%, 75% and 98%
respectively for the three farm sizes. Most of the increase is due to dimba crops. However,
income from dimba crops increases by 202% for small farms, 195 % for medium farms and
186% for large farms. The average proportion of dimba income in total income has changed
from 38% in Model I to 64% in Model II. These results highlight the importance of the
dimba land resource and the potential of intensive dimba cultivation to significantly increase
incomes.

The flip side of this result is that most labor is now absorbed in intensive cultivation
so that there is enough slack labor to generate only 21% of the total income from off-farm
IGAs compared to 49% in Model I on small farms. The increase in total income is not

enough to make up for the increase in labor use so that average returns to a person hour of

137

Table 8.4 .
Model II - Contribution of Dimba and Rainfed Land
and Off-Farm IEAs to Total Income in Non-risk Constrained Solution

 

 

Size of Farm
Small Medium Large
Rainfed Land 15 % 25 % 28 %
Dimba Land 64% 64% 67%
Off-farm IGAS 21 % l 1 % 5 %
Total Income (MK) 2,261 2,898 5,019

 

Source: Survey Data l993/LP88 Results

labor used on farm work falls from MK2.21 in Model I to MK1.56 in Model H. Selling
labor or IEAs as modelled here are not risky activities and should thus dominate the results.
However, the returns to labor from farm production are significantly higher than the mimum

wage of MKO.25, so that the use of labor on risky enterprises is optimal.

8.1.1.4 Maximum Feasible Maize Production

The non-risk constrained solution coincides with the maximum goal solution using the
maize production approach. The reason is the same as in Model I - Fertilized
maize/beans/pigeon peas is mean-minimum dominant among the rainfed maize producing
enterprises. The maximum feasible maize production goals for small and medium farms are
617 kg and 788 kg at the 0.25 probability limit (see Table 8.5). They correspond to expected
maize production levels of 1,291 kg, 3,042 kg. Large farms can achieve their maize
consumption requirements with no risk constraint. Therefore, when intensive cultivation is
practiced, small and medium farms cannot expect to achieve their minimum maize

consumption of 1,000 kg in 75 % of the cropping seasons.

for
are
MK
me:
35511

the.

(Tab

land

 

138

Table 8.5
Model 11 - Maximum Feasible Maize Production Goals
by Allowable Probability of Failure

 

 

Probability

Limit Small Medium Large
0.25 617 788 NB
0.30 800 NB NB
0.40 895 NB NB

 

NB = Not Binding, the 1,000 kg maize requirement can be met and does not constrain
solution.

Source: Survey Data 1993/LP88 Results

8.1.2 Safety First Solutions

8.1.2.1 Activities in the Solutions

The optimal crop activities were shown in Table 8.2. They are fertilized
maize/beans/groundnuts and cabbage with a small portion of chinese cabbage on dimba land
for small and medium farms. Cabbage is the sole crop for large farms. The optimal returns
are MK2,112, MK2,668 and MK5,019. They correspond to income goals of MK1,423,
MK2,067, and MK2,500. The solution for large farms is not constrained by risk. Small and
medium farms cannot hope to achieve their minimum income requirements. They can only
assure that the optimal incomes are not less than MK1,423 and MK2,067 in three quarters of

the cropping seasons.

8.]. 2. 2. Shadow Prices, Stability of Solutions and Competitiveness of
Enterprises

The shadow prices of rainfed and dimba land are MK485 and MK13,420 respectively
(Table 8.3). The safety first premium of dimba land is thus MK3,957 and that for rainfed
land MK101. These figures compare with MK1,605 and MK328 for Model I. In Model I

139

rainfed land has a higher safety first premium than in Model II. This shows that intensive
farming practices on dimba land helps in managing risk. If the farmer is very risk averse,
she should be willing to pay this risk premium over the value of crops to be produced from
the land. This value can be used in appraising irrigation investments for cases where risk is
structured as defined by yield variability in this study and farmers are as risk averse as
specified by the goals and probability levels in this model. The dual of the sufficiency
condition is ~0.51.

When safety first income goals are at their maximum feasible for small and medium
farms, hybrid maize does not present any competition to intensive maize/beans/groundnuts.
On large farms, some of the current dimba enterprises, sugarcane, maize and maize/pumpkin
leaves can be introduced at no cost. Intensive tomatoes and chinese cabbage would also cost

nothing to introduce into the optimal solution.

8.1.2.3 Contribution of Dimba Crops to Total Income
The structure of results for the contribution of various resources to income are the
same as under the non-risk constrained solutions. Dimba land contributes 60% of total

income.

8.1.2.4 The Maximum Feasible Maize Production
The expected maize production levels when there is income risk aversion are 849 kg,
2,000 kg and 4,698 kg. These are lower than in the case of no risk aversion because now

farmers need to grow those crops that maximize income while minimizing risk.

8.3 Lessons from Model II

The first lesson from Model II is that intensive cultivation can make fertilizer use
optimal even at high levels of risk aversion. This is due to the increased yields that can be

achieved with intensive use of fertilizer. Second, when the income approach is used, dimba

140
land has a higher safety first premium than in Model 1, indicating the value of this land to the
farmer in managing risk. The third lesson is that higher yields and incomes are still not
enough to make hybrid maize optimal. Thus, even when intensive dimba cultivation is

practiced farmers may find fertilized hybrid maize still too risky to adopt on rainfed land.

CHAPTER 9: MODEL III - CURRENT, INTENSIVE AND IRRIGATED
CROPPING PRACTICES

Model III allows representative farms to select crops grown with current or intensive
practices on rainfed land, and replaces dimba crops with irrigated ones. In Model III(a)
dimba crops have been replaced by irrigated horticultural crops. In Model III(b) the irrigated
crops are cereals-maize, rice and wheat. The area of irrigated land is limited to that of \
dimba land. This is done to facilitate comparison between the models.

The discussion is based on a base scenario which assumes the costs of sprinkler
irrigation and the prices that farmers attempted to get for their produce at the Mchenga and
Diamphwe projects. Sensitivity analysis is done to demonstrate the impact of lower costs and

prices on incomes.

9. 1 Base Solutions
9. 1.1 Model III(a): Irrigated Vegetables

9. 1.1 . I . Activities in Solutions

The results for Model III(a) at the three probability levels for small farms are shown
in Table 9.1. With irrigation available, yields become sufficiently stable so that differences
between risk and safety first solutions are negligible. An important difference from Model II
is that now hybrid maize can linger in the optimal solutions until the very end of the binding
but feasible region, when intensive maize/beans/groundnuts takes over completely. The
outcomes are not very different for the risk and non-risk constrained solutions: The optimal
returns differ by a few Malawi Kwachas. The major difference is in the crops on rainfed
land. The safety first constrained solution has intensive maize/beans/groundnuts while the
non-risk solution has hybrid maize/beans/pigeon peas. The safety first goal at the level of
minimum income requirement is binding only for small sized farms. For this reason and also
because the non-risk constrained solutions coincide with solutions to the maize production

approach, in most of the following discussion, reference is made only to the non-risk

141

142

constrained solutions. It is important to note that with irrigation, farmers who set their safety
goals using the maize production approach will behave in almost the same way as those who
use the income approach.

The results for the 0.25 probability limit are shown in Table 9.2. Tomatoes in
seasons 1 and 3, and cabbage in season 2 are optimal on dimba land. On rainfed land hybrid
maize/beans/pigeon peas is optimal. The maximum expected incomes are MK3,066,
MK3,947 and MK7,041 for the three farm sizes. Small farms hire labor, 220 person hours,
only in August. Medium farms hire a total of 789 person hours in six months. Large famis
hire 3,387 person hours in eight months. The level of crop selling and off-farm IEAs are
displayed in Table 9.2.

9. I . I .2 Shadow Prices, Stability of Solutions and Competitiveness of

Enterprises

The dual values of land are shown in Table 9.3. The value is MK384 for rainfed land
under non-risk constrained solution, and MK390 under the safety first solutions. For irrigated
land the values increase from MK1,665 to MK1,795 due to risk in seasons 1 and 3, and from
MK12,225 to MK13,296 in season 2.

The safety first premiums of various types of land are also displayed in Table 9.3 for
the three models. The magnitude of the safety first premium is dependent on the value of
crops grown and the dual of the sufficiency condition. Rainfed land has less safety first
premium as the risk of failure to meet the safety first goal decreases from Model I through to
Model III. Despite the comparatively high value of irrigated crops in season 2, the safety
first premium is lower than under intensive dimba cultivation because production risk is lower
with irrigation. Put differently, the incomes in Model 111 are high and stable enough that
additional units of irrigated land do not have as high a value in satisfying safety first goals as
dimba land has in Models 1 and II.

143

88: 8:882 38 :25: N85::

 

 

2.: 2.: 2.: ::.: ::v: a... 28: ::::.
2.: 2.: 2.: ::.: 8:: :2: :8: 3::
2.: 2.: 2.: ::.: ::.: :3: :8: ::.:: ::.:
2.: 2.: 2.: ::.: ::.:: ..... :8: 2:::
2.: 2.: 2.: ::.: 8:: :2 .: :8: 8::
2.: 2.: 2.: ::.: 2:: :28 :8: 8:: ::.:
2.: 2.: 2.: ::.: ::v: 28:. 8::
2.: 2.: 2.: ::.: 8:: :3: :8: :8:
2: 2.: 2.: ::.: :8: :3: :8: :2: ::.:
2.: 2.: 2.: ::.: ::.: :8: : .8
0:): 08: L:
0885 m :85
:88: ::88: :88: 88:: 5:22: 882: 888:: :8:: :8:

 

82330:: :80

 

:88:— =:8m - 9.6 .8: mmd find 883 3:338:— :<
5:89? 0885 05 98:3 :80802 3.8.: .633 98832 8::
82035 03:29:98:— :.:0:_0.H m8:D Eons—om 38m b0..:m - A3:— .0532
_.o 038.

144

R ::
:2 :2 :2
:2 8: ::: :::
:: :2
22 ::: ::: 8:
:: ::: :8 :8
:3 :2
:: K
2. :2 :2
a :2 :2 :2
:: t.
::.: 2.: 2.: 2.:
::.: 2.: 2.: 2.:
::.: 2.: 2.: 2.:
::.: 8.: ::.: ::.:
E. 2.: i.
::.: 8.2 ::.:
::.: ::.: 28.: :8:
588288 5882008 L28: .2008
0:3 52:02 :2:

...B.EOU

:038000Q
.5383: Z
83800
:0380E0m
8&3:
:3».

08:.

82

:8:.
38:2
883$
8:88:

Gun—On— GOmhonmv 2mm ghubo

:28:
8.8.:
:88:
3:: 888
582.
88,8

3:: 823:3: 80:02:05 Q80
0:): 2:85 38:8

33:2

 

mucus—om 8 :0_:_>=0< no .034

:83 323305 mad 05 .<
30:08:? 08005 05 8::: 280802 38::— ._0Bo.._ £00033. “0::
8:20:20 0:23:38:— :.:0:_0._. £5 muons—om :BE 38:“: - 3:— ~03:
Na 03:._.

 

«U. =~OrUv N .0 D~C~flh

338M wwmflmae Sun >025 ”350m

muons—om “Em bomam - HmmEm B
80:28 585% conga—gem 382 mm 063. 2a - Eons—om 8593800 35182 - ZOOMZ \a

 

145

as. M: 598089
MN in nu- I: Hung—962
:m 2 “Begum
63 v8 8N o8 3&3
SN :3
8v 8 $2
5 m: E?
a: 532
Ev a 523.
Amazonucoaomv womum>€o< .593 8::
cm: 3.3 $3. an? «momsfiu
M3,: 9% NE Na. 3585
E «meow
am a: 2. K 23m 35am
3.. I: mo in 35552sz
98 8 9. 93m :8?
Mac SN mm om 88m
3: ~35 mm a? 0%: Essa

Aaaaofic 823.2 ”5:2“

8288 Na 05¢

 

146

Table 9.3
Safety First Premium of Rainfed, Dimba and Irrigated Land in Models 1, II and III

 

 

Model I Model [1 Model 111
DV SF DV SF DV SF
NRCON Prem NRCON Prem NRCON Prem
Rainfed Land 192 328 384 101 384 6
Dimba Land 1,632 1,605 13,420 3,957 na na
IRLDSl na na na na 1,665 194
IRLDSZ na na na na 12 ,225 1 ,071
IRLDS3 na na na na 1,665 194

 

na = not applicable

DV NRCON = Dual Value for Non-risk Constrained Solution

SF Prem = Safety First Premium, the difference between dual value of non-risk constrained
solution and safety first solution

IRLDSl, IRLDSZ, IRLD83 = irrigated land seasons 1, 2 AND 3

Source: Survey Data/LP88 Results

The dual value of the sufficiency condition under Model III is very low, -0.08,
because the condition is not very constraining. Reducing the goal by MKl .00 will increase
the returns by only 8 tambala. Put differently, safety first behavior is barely necessary when
farmers use some formal irrigation. A caveat in this regard is that yields used in the analysis
do vary by season but not by year. If yearly variation of seasonal yields is allowed the
’safety first’ premium would be higher.2 Some crops compete at low cost with those in
solution: intensive maize/beans/groundnuts competes with hybrid maize/beans/pigeon peas on
rainfed land at cost of MKl llha; and onions compete with cabbage in season 2 at a cost of

(MKO.5)lha. In season 1 and 3 there is no competition with tomatoes.

 

2 A needed qualification is that sometimes other kinds of risk are introduced by
participating in irrigation. Production becomes dependent on inputs whose supply may be
unreliable.

147

Price stability analysis shows that pumpkin leaves would enter solution if its price rose
to MKO.67/ kg, in seasons 1 and 3. However, it is unlikely that in season 3 prices would go
up because markets for pumpkin leaves do glut. The price of beans would have to go up
50% to MK1.57 for that enterprise to enter solution, and the same for tomatoes in season 2.
The case for tomatoes is more likely. Thus, farmers can split their plots between tomatoes
and cabbage in season 2 if the cabbage price is anywhere near Mk1 .SO/kg, making three
possible crops with onions as shown above. In season 1, both pumpkin leaves and tomatoes
can be planted. In season 3, tomatoes are dominant. This is a season of extreme scarcity of
tomatoes due to pests and diseases. Careful management to ensure high yields can lead to
very high returns. Reports from irrigation projects indicate that extension staff and farmers

have been slow in responding to some tomato diseases [Lilongwe ADD 1993(a)].

9.1.1. 3 Contribution of Irrigated Crops to Total Income

The contribution of irrigated cr0ps to income is shown in Table 9.4. On average it is
76%. Noticeable is that the possible contribution of ojffarm IEAs goes down to 14% from
49% in Model I for small farms, 3] % to 8% for medium farms and 12 % to 2% in the case of
large farms. This is evidence that irrigation can absorb a lot of labor. Our approach to off-
farm IEAs is to assume alternative use of agricultural slack labor at MKO.25. The
opportunity to use it may not exist, and in that instance irrigation provides a way of assuring

that there is less under- or unemployment.

148

Table 9.4
Model III(a) - Contribution of Irrigated and Rainfed Land
and Off-Farm IEAs to Total Income

 

Size of Farm

Small Medium Large
Rainfed Land 10 % 17 % 20 %
Irrigated Land 76 % 75 % 78 %
Off-Farm IEAs 14 % 8 % 2 %
Total Income (MK) 3,066 3 ,947 7,041

 

Source: Survey Data 1993/LP88 Results

As in the cases of Models I and II, the contribution of land depends on the
profitability of the enterprises grown on it. In Model III an important determinant of
profitability is the cost of irrigation. The lower the cost of irrigation the higher the
contribution of irrigated land to income. In particular, gravity fed/furrows irrigation can lead

to farms generating up to 81% of income from their irrigated land.

9.1.1.4 Maximum Feasible Maize Production

The maximum feasible maize yields are the same as under Model 11; the reason being
that the same optimal activity hybrid maize/beans/pigeon peas will give these yields.
However the total income possible does change and is the same as in the non-risk constrained

solutions for Model III.

9.1.2. Sensitivity Analysis for Model III(a)
9.1.2.1 Sensitivity to Changes in Cost of Irrigation

The results of expected income by type of irrigation are presented in Table 9.5 and
provide some important insights. First, the returns to farmers for the four types of irrigation

show the profit that farmers can generate if government were to subsidize sprinkler irrigation.

149

Second, even seemingly high cost sprinkler irrigation has the potential to give returns above

the costs of capital and operations. This is in spite of low yields for some seasons.

Table 9.5
Model 111 - Returns Under Various Types of Irrigation for Small Famis

 

Costs Relative

 

Pump/Sprinkler
Costs (MK)‘ System Returns (MK)
Pump/Sprinkler 7,379 100% 3,066
Pump/Furrows 4,776 34% 3,386
Borehole/Furrows 2,166 70% 3,726
Tank/Furrows 1,852 74% 3,767
Gravity/Furrows 621 91 % 3,927

 

1/ Costs include amortized debt, maintenance and operating costs

Source: Department of Irrigation, Lilongwe ADD & LP Analysis

9.1.2.2 Sensitivity to Changes in Prices of Output (With Sprinkler Irrigation)

Sensitivity analysis for the price of tomatoes, and chinese cabbage. The prices of
these two crops vary much by season and they are the most difficult to sell because of high
perishability. When the price of tomato in seasons 1 and 3 is reduced from MK2.00 to
MKl.00, the returns drop to MK2,691 for small farms, and chinese cabbage comes into
solution in both seasons. The dual values of irrigated land in seasons 1 and 3 fall accordingly
to MK216. Then pumpkin leaves show strong competition. The price increase necessary for
pumpkin leaves to enter solution is only MK0.04. The same is true for beans, at MKO. 10.
Tomatoes could go back into solution at MK1.41. Therefore, if farmers cannot get a good
price for tomatoes in seasons 1 and 3 ( i.e. > ‘MK1.41) they can also plant pumpkin leaves
and beans.

Holding the above scenario constant, and adding a lower price of chinese cabbage

leads to a lower income still, MK2635, since no activity can substitute the two chinese

150

cabbage crops and the land should be left fallow. This result is of special concern since leafy
vegetables are difficult to market (LADD 1993). Farmers may have to reckon with low
prices and it may be optimal to not grow crops in some seasons. However, they can grow
crops if less costly irrigation is used. Cabbage still remains in solution under this scenario.
However, farmers will need to use earnings from growing cabbage to pay the whole year’s
debt. Since the debt is split into MK313 for every season, MK626 will be subtracted farther
from MK2,635 to give MK2,009. In this case, it is still possible for farmers to make a
positive yearly income. They are at almost the same level as in Model 11 (MK2112)3. If
irrigation is done by farmers without dimba land and with only 0.45 hectare of rainfed land,
this income represents a significant improvement since dimba land small sized farms can
expect an income of only MK946. This income improves the situation of small sized farms

with dimba land from MKl,38O in Model I to MK2,009 in Model III(a).

9.1.2.3 Sensitivity to Changes in Wages

The minimum wage of MKO.25 is reduced by 50% for both labor selling and labor
hiring activities. The results show that small farms’ expected returns fall by MK30 and for
large farms they increase by MK103 to MK7144. Medium farms are not affected because

they sell as much labor as they hire.

9. 1.3 Model III(b): Irrigation of Cereals

Model III(b) deals with growing cereal crops of maize, rice and wheat using gravity
fed irrigation in addition to rainfed crops. Two cropping seasons with irrigation are modelled
in addition to the normal rainfed cultivation practices. Enterprise budgeting showed gravity
fed irrigation to be the most financially robust way to irrigate cereals. (See Tables A6.8 and

A6.9, and A6.16-A6.19 in Appendix A for gross returns to growing cereals with the different

 

3 This is the return at maximum goal in Model II. Since irrigation practically removes
risk, it is fair to compare the two figures.

151

irrigation methods.) Wheat and maize have not been grown extensively using irrigation.
Enterprise budget analysis shows that when maize is grown intercropped with beans it may be
more profitable than growing rice monocropped. The emphasis on irrigating rice has been
justified by its stronger foreign exchange earning ability compared to maize. Irrigating wheat
may increase production and thereby save foreign exchange since wheat is mostly imported.
The demand for wheat is ever-increasing, with increased urbanization and demand for fast
foods. The government of Malawi has expressed need to encourage growing of wheat with

irrigation [Malawi Government 1992(c)].

9. 1.3.1. Activities in Solution

When the representative farms are operated with no risk aversion, it is optimal to
grow hybrid maize, with beans and pigeon peas, and relay field peas on rainfed land. Wheat
is optimal in both seasons on irrigated land. The gross returns are MK1,223, MK1,536 and
MK2,361 with respect to farm size. These figures are lower than for those for Model I. The
shadow price of MK681 for the two irrigation seasons is the returns to growing wheat. Those

of rainfed land remain at MK381 and MK41.

9.1.3.2. Shadow Prices, Price Stability and Competitiveness of Solutions

Under the maize production approach, the dual values for land are MK390, MK47,
and MK766 for rainfed, relay rainfed and the two irrigation seasons respectively.
Price stability analysis shows that wheat will remain in solution unless its price drops to
MKO.39. Intensive maize/beans/groundnuts competes with hybrid maize/beans on rainfed
land. On irrigated land there is no competition to wheat from rice or maize. Forcing them
into solution would cost MK643 and MK569 respectively.

In this model, the maize safety first goals are not binding, except for small farms.
Thus, small farms can now meet cereal consumption requirements but are still constrained by

risk aversion behavior. With the income approach, intensive maize/beans and groundnuts are

 

 

152
optimal on rainfed land and wheat in the two irrigation seasons. Hybrid maize is optimal at
moderately high levels of risk aversion. Large, medium and small farms can assure that the
expected returns of MK2,333, MK1,524 and MK1,218 are above goal levels of MK41],
MK680 and MK1451 in 75% of the cropping seasons.

9.2. Lessons from Model 111

With irrigation, farmers can meet their minimum income requirements in 75 % of the
cropping seasons especially if lower cost irrigation is used. Model small forms may not be
self-suflicient in maize production if all the irrigated land is used for horticultural crops, but
they can earn enough income to meet their subsistence needs in 75 % of cropping seasons.

The representative households may now be able to use almost all their labor on their
farms. This can help to control rural-urban migration. In addition, the returns to labor used
on the farm are higher than minimum wage of MKO.25.

Farmers without dimba land can benefit from irrigation, even the costly sprinkler
system. This is the case even with low yields and prices of crops in some seasons. It is
necessary that they get good yields and prices in at least one season to profitably use the
system.

Sensitivity analysis results show that even when prices of some horticultural crops are
lowered, the optimal solutions do not include unfertilized maize crops, as was the case in
Model 1. Thus, irrigated cultivation may enable adoption of fertilizer on rainfed maize. In
Model II, hybrid maize is optimal at low levels of safety goals. As soon as the sufficiency
condition becomes binding it is forced out of solution because it aggravates the risk already
existent with dimba cultivation. In Model 111, even at moderately high safety goals, hybrid
maize remains in solution because the stable yields of irrigated crops make it less risky to
grow hybrid maize. This is good for both food and income security since hybrid maize

increases returns to rainfed land.

153

Irrigation may be considered superior to intensive dimba cultivation in that it can
encourage the adoption of fertilized hybrid maize varieties on rainfed land by farmers with
moderately high risk aversion, even when they view it as cash crop (i.e. income approach).
This is important because the growing of maize using most irrigation methods may be
unprofitable; therefore, other sources of stable income may be needed to cushion against the
risk of growing it under rainfed conditions. This can have a very significant positive impact
on food security if the hybrids are of types preferred by farmers and can also improve returns
to rainfed land.

Model III(b) has an important lesson. Although it leads to lower gross returns than
Models I and III and III(a), it has potential to assure representative farms of all sizes of their
cereal requirements in 75% of the cropping seasons. In Models I, II and III(a) only large
farms could hope to achieve cereal self-sufficiency in 75 % of the cropping seasons. In
situations where markets for cereals do not function well and markets for horticultural crops
are uncertain, this may be the most attractive way of dealing with food insecurity in
households. In this case, rather than just growing only rice, households should be helped to

adopt mixed cropping with irrigation.

 

 

CHAPTER 10: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

10.1 Summary of Objectives and Research Methodology

The Government of Malawi places considerable emphasis on ensuring that all people
have a timely, reliable and nutritionally adequate food supply. Yet food insecurity is still a
major problem among rural smallholder households. Under existing agricultural polices, the
chances of most households attaining food security are dismal owing to the risky nature of
rainfed, low-input crop farming.

The goal of this study was to investigate the potential of intensive dimba cultivation
and small scale irrigation to improve the food security position of smallholder households.
The specific objectives were:

1) to understand the current pattern of land use of both dimba land and rainfed land for
agricultural production in the Chiradzulu North Extension Planing Area (EPA);

2) to compile enterprise budgets for farm activities congruent with land and labor use as
determined above;

3) to estimate the contributions of dimba crops to household food income and food
security;

4) to identify technically feasible crop/irrigation technology packages that can be used to

increase production of crops;

5) to estimate the possible contributions of irrigated crops to household income and food
security;

6) to select crop/ irrigation technology packages that contribute most to improving food
security;

7) to make policy and research recommendations based on the findings of the study to
appropriate institutions.
To achieve these objectives, a field study was conducted during November 1992 to

February 1993 among smallholder farming households in Chiradzulu North Extension

154

155

Planning Area of Blantyre ADD. Sampling of households was done purposively to get
enough households with dimba land. Three questionnaires (one for households, one for
enterprises, and one for garden measurement) were used to obtain data consistent with the
objectives. Secondary data sources were used to supplement survey data to get yields on
irrigated crops and the kind of techniques that can be used for irrigation, as well as farmers’
experience with irrigation.

The data are used to describe land and labor use patterns in the area and to compile
enterprise budgets to supply technical co-efficients that are used to implement a safety first
Target MOTAD programming model. The use of this model assumes that farmers are risk
averse and practice safety first behavior when deciding what crop production or IEAs to
engage in. The safety first behavior is modeled using Telser’s probabilistic criterion. The
criterion is imposed in Target MOTAD with a sufficiency condition that uses Atwood’s
(1985) lower partial moments approach. The algorithm selects the activity mix which
optimizes expected income while simultaneously enforcing probabilistic constraints upon
negative deviations of specified goal variables from a target level using the least constraining
linear lower partial moment stochastic inequality. Two different probablistic constraint
approaches are specified, one using maize production and the other using income.

To implement the Target MOTAD, a programming tableau of activities and resources
constraints was formulated. The activities included were crop production and selling
activities, labor hiring and selling activities, and deviation counters. Constraints were family
labor, land, accounting rows, and chance and deviation rows. The technical co-efficients

were derived from enterprise budgets of current, intensive and irrigated crops.

10.2 Summary of Farm Survey Results
The sample size was 270 of which nearly 20% were female headed households.
Fifty- two percent of the household heads were literate. Median household size was 5. Small

farms (less than 0.7 ha), medium farms (< 0.7 up to 1.5 ha), and large farms (1.5 or more

156

ha) made up 16.3%, 36.3% and 47.4% of the sample respectively. Due to selection of
households possessing dimba land, the households in the survey were generally better
endowed with land than most in the area with a mean land holding of 1.7 hectares. (The
average in the area is estimated at 0.8 ha.) On average, 1.5 ha was rainfed land and 0.24 ha
dimba land. Small farms had an average of 0.45 ha of rainfed land and 0.13 ha dimba land;
for medium farms the respective figures were 1.06 ha and 0.17 ha, and for large farms 2.49
ha and 0.33 ha.

The dominant crop mixtures grown on rainfed land were maize/beans, peas, maize
alone, maize/beans/pigeon peas, and maize/beans/groundnuts. For dimba land they were
sugarcane, maize, cabbage and tomato.

The survey households were more likely to use fertilizer than is the case among
typical smallholder households in Malawi. A large minority hired labor for fertilized maize-
based crop mixtures grown on rainfed land. Pesticides were used on dimba crops of chinese
cabbage, cabbage, tomato and rape.

Although all households farmed some land, nearly 30% of heads of households do not
consider farming their main occupation. They generally produce enough maize when rainfall
comes in good amounts. The majority, 71%, are willing to use and invest in irrigation.

However, they foresee problems in organization and implementation.

10.3 Summary of Mathematical Programming Results
10. 3. 1 Cropping Activities

The results of Model I for base scenario and sensitivity analysis show that in the
farmers’ current situations, some risky crops (e. g. fertilized maize/beans/pigeon peas) are
optimal when sources of income from dimba crops are high and stable, e.g. from sugarcane,
or when safety first goals are based on maize production. Not fertilizing maize-based crop

mixtures can be optimal when risk avoidance behavior is very conservative (i.e. safety first

157
goals are very high and/or allowable probability of failure is low) and income from dimba
crops is unstable and very low in drought years.

This result indicates that to encourage farmers to use fertilizer on maize, maize yields
must be stabilized (as by irrigation) or farmers need to have other non-rainfed-land stable
sources of income to manage maize production risk. This is borne out by observance that it
is farmers who have collateral or other sources of income that usually participate in credit
programs. for maize (Center for Social Research 1988, Ngwira 1988).

The results of Model 11 show that medium intensity levels of fertilizer use, as in
intensive local maize/beans/groundnuts, may be optimal even at high levels of risk aversion.
This is made possible by the higher incomes from intensive dimba cultivation. However, the
high level of fertilizer use as on hybrid maize is still too risky for farmers to adopt.

With irrigation in Model III(a), hybrid maize is optimal on rainfed land with no risk
aversion in both versions. With risk aversion, stable incomes from irrigated crops make
intensive maize/beans lgroundnuts optimal at the highest levels of risk aversion; and hybrid
maize is optimal at moderately high levels of risk aversion.

In Model III(b), hybrid maize is also optimal at moderately high levels of risk
aversion. Wheat is optimal in season one, and maize/beans in season two. The results of
Model HI supports the conclusion that both stable and high incomes from non-rainfed land

cultivation should encourage the use of fertilizer on maize based crop mixtures.

10. 3. 2. Expected Income and Contribution of Dimba Crops to Total Income

The results from the three models are summarized in Table 10.1. They include

expected income, maximum feasible goals and optimal levels of maize production.

10. 3. 2. 1. Current Cropping Practices

The returns to production on representative farms in the current situation are very

low: MK1,380, MK1,657 and MK2,529 for small, medium and large farms. With them,

 

158

small and medium farms achieve 55 % , and 66% of the minimum income requirements of
MKZ,500. Only large farm households can be certain of achieving these minimum income
requirements in 75 % of the cropping seasons. With risk aversion, the expected incomes drop
to MK1,359, MK1,638 and MK2,471, and these incomes can be greater than the income goal
levels of MK576, MK903 and MK1,863 in 75% of cropping seasons. These results suggest
that many households in the small and medium size categories are affected by poverty
resulting in malnutrition, morbidity and mortality.

Dimba land contributes significantly to the income: 35%, 38% and 47 % of total
income on small, medium and large farms in the base scenario. Thus, under current cropping
practices, dimba land can be of equal or greater importance compared to rainfed land for
providing livelihoods to households depending on the size of land holdings. The level of
contribution will also depend on the value of crops grown which in turn depends on costs and
prices of input and output. However, under most scenarios of low prices, dimba land still
contributes a significant portion of total income. This is important given the small sizes of
these dimba plots. It is evident that creating small tracts of irrigated land has the potential to
significantly change the food security situation of households through providing higher
incomes.

Under current circumstances, model farms can generate a good portion of their
income from off-farm income earning activities (IEAs) especially in the small farm category
where 49% of income can be so derived. These figures are 31% and 12% for the medium
and large farms. The model small farms have underemployed agricultural labor in almost all
months, totalling 2,704 person hours which can be used in off-farm IEAs. Medium sized
farms have 2,055 person hours. The large farms need to hire labor in some months (totalling

855 person-hours) and have 1,220 person hours to put to IEAs or employment at other times.

 

159
10. 3. 2. 2 Intensive Cropping Practices

When enterprises that use intensive fertilizer and chemical input levels are considered,
potential expected incomes increase significantly. For farmers indifferent to risk, incomes for
small farms increase from MK1,380 in Model I to MK2,261 in Model II, for medium farms
from MK1,657 to MK2,898, and for large farms from MK2,529 to MK5,019, meeting 62%
and 80% of the requirements of small and medium farms. However, small and medium
farms can not be certain of achieving incomes higher than subsistence requirements in 75 % of
cropping seasons. Given the land resources, intensive crop practices without irrigation may
not solve the food security problems of small and medium farm families.

What changes significantly when intensive enterprises are considered is that dimba
crops can contribute much more to income than rainfed crops because of the very high yields
that can be achieved. The portion of income that can be derived from dimba crops increases
from 35% in Model I to 64% in Model II for small farms; from 38% to 64% for medium
farms and from 47% to 67% for large farms.

Due to the high labor demand of intensive crops, slack labor available for off-farm
IEAs drops. Consequently, income that can be generated from IEAs falls from 49% in Model
Ito 21% in Model 11 for small farms. For medium farms it falls from 31% to 11%, and for

large farms the drop is from 12% to 5%.

10. 3. 2.3 Irrigated Crop Practices

With sprinkler irrigation for horticultural crops, potential incomes increase to
MK3,066, MK3,947 and MK7,041 respectively for small, medium, and large farms.
However, it is possible to get MK3,927, MK5,072 and MK9,225 when gravity-fed irrigation
is used for horticultural crops. Thus, it is possible for all farm sizes to meet minimum
income requirements with irrigation for horticultural crops. Irrigated land can contribute

from 75 % to 81% of expected returns depending on the type of irrigation that is used.

160

With sprinkler irrigation for horticultural crops, all the representative farm sizes need
to hire labor-220, 789 and 3,387 person-hours with respect to small, medium, and large farm
sizes respectively. The proportion of income that comes from off-farm IEAS is 14% , 8% and
2% respectively. This is a significant drop from 49%, 31% and 12% in Model 1. Thus,
irrigation has the capacity to absorb slack labor and provide higher returns to labor than
minimum wage employment.

When gravity fed irrigation is used for cereals, income are lower--MK1,223,

MK1,536 and MK2,361 with respect to small, medium and large farms.

 

161

Table 10.1
Comparison of Expected Returns, Maximum Feasible Goals and Maize Production
Models I, II and III

 

 

NRCON SAFETY FIRST
Exp.Income' Exp.Income(MK) Exp.Income(MK)
(MK) (Goal- 2 (Goal-Maize Kg)2
Maize Prod. (Kg) Maize Prod. (Kg)
Model 1
Small Farms 1380 1359 (576) 310 1380 (300) 668
Medium Farms 1657 1638 (903) 1251 1657 (708) 1480
Large Farms 2529 2471 (1864) 2779 2429 (1000)
3476
Model 11
Small Farms 2261 2112 (1423) 849 2261 (617) 1292
Medium Farms 2898 2668 (2067) 2000 2898 (788) 3042
Large Farms 5019 5019 (2500) 2898 5019 (1000) 3478
Model III(a)
Small Farms 3066 3061 (2500) 853 3066 (617) 1292
Medium Farms 3947 3847 (2500) 3042 3947 (788)
3042
Large Farms 7041 7041 (2500) 7146 7041 (1000) 7146
Model HI(b)
Small Farms 1223 1291 1218 (411) 855 1150 (1000) 1675
Medium Farms 1536 1524 (680) 2004 1536 (1000) 3042
Large Farms 2361 2333 (1451) 4699 2361 (1000) 7146

 

1/ Maize produced under non risk constrained solution is same as under maize
production approach except for small farms in Model III(b). The goal for NRCON

solutions is zero.

2/ The expected income can be achieved while meeting the goal in 75% of cropping

86880118 .

3/ Model I-Current Crop Practices; Model II-Intensive Crop Practices; Model III(a)
Sprinkler Irrigation for Horticultural Crops; Model III(b) - Gravity-fed Irrigation for

Cereals .

NRCON : Non-risk constrained solutions

Safety first:

Safety first solutions

 

 

162

 

10. 3. 4. Shadow Prices of Resources and Constraints

Under Model I, dimba land has a higher safety first premium than that of irrigated
land in season 2 when the dual value is many times higher. This reflects the high value of
land to meeting the sufficiency condition in Model 1 when income safety goals are very high.
Put differently, under current crop production practices farmers would greatly benefit from

acquiring more irrigated land to enable them to achieve safety first income goals as specified

in this model.

10. 3. 5. Maximum Feasible Maize Production

Under current crop production practices, only large model farms can meet the
specified minimum maize requirement of 1,000 kilograms per household in at least 75 % of
the cropping seasons from own production. Medium farms can be self-sufficient in maize in
at least 67% of the cropping seasons with intensive cultivation practices. Small farms cannot
expect to be self-sufficient in maize at the probability constraints used here even with

irrigation, if irrigated land is not used to grow cereals for food. However, they can generate

enough income to meet their subsistence income requirements.

10.4. Lessons from Analysis of Programming Results

1) Current and intensive crop farming may not lead to returns or maize yields that solve
the food security problems of small and medium farms. Irrigated crop farming has
potential to increase income of small farms or help them attain maize self-sufficiency.
This finding indicates that for farms without dimba land, food insecurity can be dealt
with by irrigating either horticultural or cereal crops depending on market conditions

for horticultural crops.

2) Intensive dimba cultivation generates high enough incomes to cushion farmers from

the risk of fertilizing local maize varieties or using fertilizer at medium intensity. It

may thus be concluded that if employment and IEAs are limited, and irrigation is not

3)

4)

5)

6)

163
feasible, intensifying dimba cultivation is not just an alternative but should be part of
the strategy to improve food security.
When income safety goals are high, hybrid maize is not optimal and cannot compete
with other maize based crop mixtures. This result has been supported by other
studies which show that it is not worthwhile to grow hybrid maize in bad years
(Smale 1991). However, this study goes further by showing that as long as farmers
are very risk averse (i.e. want to meet income subsistence needs in 75 % of cropping
seasons), and bad years come as frequently as once every four years, farmers should
not grow hybrid maize. However, if they have irrigation, they may grow hybrids or
use fertilizer intensively even when they have moderately high risk aversion.
All four models do not include hybrid maize at the highest feasible income safety
goals. Thus, as long as farmers a) view hybrids as cash crops and not a source of
household food consumption requirements, and b) are very risk averse, they may not
grow hybrid maize unless its price or on-fann yield increases significantly. This
finding supports the view that varietal change in breeding and extension should
encourage the new trend which emphasizes semi-flint varieties, e. g. MH18 if farmers
are to adopt hybrids. Although growing hybrids does not generate the same income
as dimba crops, it increases the income returns to rainfed land by 96% , and also
assures higher minimum maize production. Both of these contribute positively to food
security.
The LP. solutions based on the maize production approach have maize-based crop
mixtures using fertilizer. If farmers set their safety first goals using the maize
production approach, failure of to use fertilizer may be construed as lack of access to
fertilizer.
When gravity fed irrigation is used for cereals, model farms achieve lower incomes,

but their cereal consumption requirements can be met. Thus, if it is not possible to

establish marketing arrangements that make growing vegetables profitable, gravity fed

 

164
irrigation should be used to grow cereals. This should be the case for most rural
areas that are far from urban centers. When farmers in the small and medium
landholdings set their safety first goals using the income approach, irrigating cereals is

not attractive because the expected incomes cannot meet their subsistence

requirements.

7) Most of the risk—constrained solutions show the need for diversification. Different
crops do well in different years. Diversification is shown to be important to optimize
income in situations of risk aversion and also to assure food security. It is important

that credit and extension which is currently centered on maize be diversified.

10.5 Recommendations for Dimba Cultivation and Irrigation

Having found that 1) smallholder farmers should not grow hybrid maize if they are
very risk averse and income oriented in their safety goals, 2) moderately risk averse farmers
can grow hybrid maize with gravity-fed irrigation, 3) small farms will not meet their goals of
achieving maize consumption requirements in 75 % of the cropping seasons unless they

irrigate cereals, and 4) intensive dimba cultivation and irrigation has significant potential to

increase income it is recommended that:
smallholder farmers with dimba land should be targeted with credit and extension

1)
for dimba crops, rather than rainfed maize, particularly those in the smallest size
category.

2) farmers without dimba land should be assisted in establishing and investing in

irrigation for either horticultural or cereal crops, depending on the market situation.
The government recognizes that small farms can not produce their food requirements

from their gardens or meet minimum income requirements. The results of this study indicate
that irrigation can enable households with small farms to generate enough income to purchase

food. Irrigation can have the double-edged role in food security of increasing self-sufficiency

in maize and generating adequate income to purchase food.

 

 

165

Having found that as long as farmers are risk averse and view hybrid maize as a cash
craps (i.e. income approach), and not a source of household food consumption requirements,
they may not grow it unless its price or on-farm yield increases significantly, it is
recommended that:

3) maize breeding research and extension should encourage the new trend

which emphasizes semi-flint varieties.

These semi-flint varieties have storage and processing characteristics preferred by
farmers for own consumption maize. Although growing hybrids does not generate as much
income as dimba crops, it increases the income returns to rainfed land by 96 % and assures a
higher minimum maize production.

Having shown that when farmers set their safety goals using a maize production
approach, the failure to use fertilizer can be construed as lack of access to fertilizer, and
cognizant of evidence that many smallholder farmers do not use fertilizer, it is recommended

that:
4) the credit and extension program should allow wider access to inputs.
This may include detaching credit from government extension programs and
encouraging formation agricultural credit associations.
Since safety first solutions show the need for diversification and since diversification
spreads income risk, and enhances food security, it is recommended that:
5) the credit and extension which are currently centered on maize be diversified
to put more emphasis on other crops.

This should help to make it less risky to fertilize maize.

1 0.6 Recommendations for Future Research

For future research, it is recommended:

6) more farm production studies of horticultural craps be done to enhance the
quality of input-output data.

 

166
This would certainly improve the reliability of subsequent studies validating the results

presented here.
marketing studies be conducted on horticultural crops.

7)

This is necessary in order to improve understanding of vegetable markets and the
possible ways to improve their efficiency and assess the price to be expected with additional
products from different levels of irrigated land. As optimal crop/technology packages depend
on marketing and pricing arrangements, these studies can help in refining decisions on crops
that should be grown on dimba in each season and on rainfed land. This will have
implications for government infrastructural investments.

community and institutional organization studies be conducted for local

management and financing of irrigation projects.
The information would help increase the chance of successfully implementing

8)

irrigation projects.
All in all, this study provides ample evidence that intensive dimba cultivation and

small scale irrigation can have a two-fold enhancing effect on household food security by both

encouraging adoption of fertilized maize and generating significantly higher incomes with

which to purchase food.

 

 

 

APPENDIX A

 

EXPLANATORY NOTES FOR ABBREVIATIONS USED IN ENTERPRISE

maint. of farm
equip.

inv. in irrig.
equip.

1/prep.
irrig.
fert.

harv.

BUDGETS

tin containing seed and volume of 300 ml

kilogram

measuring cup with volume of 50 ml

Malawi Kwacha (at time of study US$1 = MK4.20)

person hours of labor

packet, the packet for chemicals applied to tomato and cabbage
weighs 200 g

maintenance of farm equipment
investment in irrigation equipment

land preparation
irrigate
fertilize

harvest

167

 

 

168

Table A6.]
Enterprise Budget for 1 Hectare of Current Non-Fertilized Local Maize

 

 

Quantity Price Value

(MK) (MK)

Output (kg) 448.00 0.33 161.50
Variable Costs

Seed (kg) 25.00 0.33 8.25

Fertilizer (kg) 0.00 0.00 0.00

Marketing (p/h) 488.00 0.03 12.20

Maint. of Farm Equip. 0.33 19.85 6.55

Total Variable Costs 27.00

Gross Margin 134.50

Labor Input (Median) Person-hours

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 85
February 0
March 0
April 25 25
May 25 25
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 85
TOTAL ' 302 38 170 50 560

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

169

 

 

Table A6.2
Enterprise Budget for l Hectare of Current Fertilized Local Maize

Quantity Price Value

(MK) (MK)

Output (kg) 1,051.00 0.33 346.83
Variable Costs

Seed (kg) 25.00 0.33 8.25

Fertilizer (kg) 110.00 1.00 110.00

Marketing (p/h) 1,051.00 0.03 26.28

Maint. of Farm Equip. 0.33 19.85 6.55

Total Variable Costs 151.00

195.33 "

GrossMargin

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 28 113
February 0
March 0
April 38 38
May 38 38
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 28 113
TOTAL 302 38 170 56 76 642

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

170

Table A6.3
Enterprise Budget for 1 Hectare of Intensive Local Maize

 

 

Quantity Price Value

(MK) (MK)

Output (kg) 2,561.00 0.33 830.50
Variable Costs

Seed (kg) 25.00 0.33 8.25

Fertilizer (kg) 160.00 1.00 160.00

Marketing (p/h) 2,516.00 0.03 62.90

Maint. of Farm Equip. 0.33 19.85 6.55

Total Variable Costs 237.70

Gross Margin 592.80

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 41 126
February 0
March ‘ 0
April 91 91
May 91 91
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 41 126
TOTAL 302 38 170 82 182 774

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

171

Table A6.4

Enterprise Budget for l Hectare of Hybrid Maize

 

Quantity Price
(MK)
Output (kg) 3,652.00 0.33
Variable Costs
Seed (kg) 25.00 0.33
Fertilizer (kg) 370.00 1.00
Marketing (p/h) 3,652.00 0.03
Maint. of Farm Equip. 0.33 19.85
Total Variable Costs
Gross Margin
Labor Input (Median) Person-hours
Month L/Prep Plant Weed Irrig Fert Harv
January 85 41
February
March
April 91
May 91
June
July
August 151
September 151
October 19
November 19
December 85 41
TOTAL 302 38 170 82 182

Value
(MK)

1,205.16

Total

126

91
91

151
151
19
19
126

774

8.25
370.00
74.98
6.55

459.78

745.38

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

172

Table A6.5

Enterprise Budget for 1 Hectare of Current Non-Fertilized Maize/Beans

 

Quantity
Output (kg)
Maize 514.00
Beans 45.00
Gross Value
Variable Costs
Seed (kg)
Maize 25.00
Beans 40.00
Marketing (p/h) 559.00
Maint. of Farm Equip. 0.33
Total Variable Costs
Gross Margin

Labor Input (Median) Person-hours

Month

January 85
February

March

April

May

June

July

August 151

September 151

October 19
November 19
December 85

TOTAL 302 38 170 0

Price
(MK)

0.33
0.65

0.33
0.65
0.03
19.85

L/Prep Plant Weed Irrig Fert Harv

27
27

54

Total

85

27
27

151
151
19
19
85

564

Value
(MK)

169.62
29.50

415.77

8.25
26.00
13.97

6.55

54.77

361.00

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

173

Table A6.6

Enterprise Budget for l Hectare of Fertilized Local Maize/Beans

 

370.62

8.25

Price
(MK)
Output (kg)
Maize 0.33
Beans 0.65
Gross Value (MK)
Variable Costs
Seed (kg)
Maize 0.33
Beans 0.65
Fertilizer (kg) 1.00
Marketing (p/h) 0.03
Maint. of Farm Equip. 19.85
Total Variable Costs
Gross Margin
Labor Input (Median) Person-hours
Month L/Prep Plant Weed Irrig Fert Harv
January 26
February
March
April 48
May 48
June
July
August 151
September 151
October 19
November 19
December 26
TOTAL 302 38 170 52 96

Total

111

48
48

15.1
151
19
19
111

658

Value
(MK)

44.80

415.42

26.00
120.00
29.83
6.55

190.63

224.79

 

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

174

Table A6.7
Enterprise Budget for l Hectare of Intensive Fertilized Maize/Beans

 

 

Quantity Price Value
(MK) (MK)
Outpm (kg)
Maize 2,049.00 0.33 676.17
Beans 151.00 0.65 98.15
Gross Value (MK) 774.32
Variable Costs
Seed
Maize (kg) 25.00 0.33 8.25 ..
Beans 40.00 0.65 26.00
Fertilizer 160.00 1.00 160.00
Marketing (p/h) 2,200.00 0.03 55.00
Maint. of Farm Equip. 0.33 19.85 6.55
Total Variable Costs 255.86
Gross Margin 518.45

Labor Input (Median) Person-hours

 

Month LlPrep Plant Weed Irrig Fert Harv Total
January 85 34 119
February 0
March 0
April 56 56
May 56 56
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 34 119
TOTAL 302 38 170 0 68 112 690

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

175

Table A6.8

Enterprise Budget for 1 Hectare of Hybrid Maize/Beans Irrigated Using Gravity/Furrows

 

Quantity Price
(MK)
Output (kg)
Maize 2,999.00 0.33
Beans 402.00 0.65
Gross value 1,
Variable Costs
Seed (kg)
Maize 25.00 0.330
Beans 40.00 0.650
Fertilizer (kg) 370.00 1.000
Marketing (p/h) 3,401.00 0.025
Maint. of Farm Equip. 0.33 19.850

Maint. of Irrig. Equip.
Investment in Irrig. Equip.

Total Variable Costs
Gross Margin

Labor Input (Median) Person-hours

Month L/Prep Plant Weed Irrig Fert Harv Total
January

February

March

April 151 151
May 151 38 150 339
June 85 150 84 319
July 85 150 84 319
August 298 298
September

October

November

December

TOTAL 302 38 170 450 168 298 1,426

Value
(MK)

989.76
261.30

250.97

8.25
26.00
370.00
85.03
6.55
50.00
259.00

804.83

446.14

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

176

 

Table A6.9
Enterprise Budget for 1 Hectare of Hybrid Maize/Beans Irrigated Using
Pinup/Sprinklers
Quantity Price Value
(MK) (MK)
Maize (kg) 2,999.00 0.33 989.76
Beans (kg) 402.00 0.65 261.30
Gross value 1,250.97
Variable Costs
Seed (kg)
Maize 25.00 0.330 8.25
Beans 40.00 0.650 26.00
Fertilizer (kg) 370.00 1.000 370.00
Marketing (p/h) 3,401.00 0.025 85.03
Maint. of Farm Equip. 0.33 19.850 6.55
Fuel and Oil 558.00
Maint. of Irrig. Equip. 450.00
Inv. in Irrig. Equip 2,332.00
Total Variable Costs 3,835.83
Gross Margin -2,584.86
Labor Input (Median) Person-hours
Month L/Prep Plant Weed Irrig Fert Harv Total
January
February
March
April 151 151
May 151 38 150 339
June 85 150 84 319
July 85 150 84 319
August 298 298
September
October
November
December
TOTAL 302 38 170 450 168 298 1 ,426

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

177

 

Table A6.10
Enterprise Budget for 1 Hectare of Current Non-Fertilized Local Maize/Beans/
Pigeon Peas
Quantity Price Value
(MK) (MK)
Output (kg)
Maize 989.00 0.330 326.37
Beans 97.00 0.650 63.05
Pigeon Peas 76.00 0.400 30.40
Gross value 419.82
Variable Costs
Seed (kg)
Maize 25.00 0.330 8.25
Beans 40.00 0.650 26.00
P/Peas 6.00 0.400 2.40
Marketing (p/h) 1,162.00 0.025 29.05
Maint. of Farm Equip. 0.33 19.850 6.55
Total Variable Costs 72.25
Gross Margin 347.57

Labor Input (Median) Person-hours

Month

January
February
March
April

May

June

July
August
September
October
November
December

TOTAL 302

151
151

L/Prep Plant Weed Irrig Fert Harv
85
38
38
19
19
85
38 170 0 0 76

Total

85

38
38

151
151
19
19
85

586

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

178

Table A6.11
Enterprise Budget for 1 hectare of Current Fertilized Local Maize/Beans/Pigeon Peas

 

 

Quantity Price Value
(MK) (MK)
Output (k8)
Maize 1,459.00 0.330 481.47
Beans 111.00 0.650 72.15
Pigeon Peas 72.00 0.400 28.80
Gross value 582.42
Variable Costs
Seed (kg) ‘
Maize 25.00 0.330 8.25
Beans 40.00 0.650 26.00
P/Peas 6.00 0.400 2.40
Fertilizer (kg) 136.00 1.000 136.00
Marketing (p/h) 1,642.00 0.025 41.05
Maint. of Farm Equip. 0.33 19.850 6.55
Total Variable Costs 220.25
Gross Margin - 362.17

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 3 l 1 16
February 0
March 0
April 68 68
May 68 68
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 31 116
TOTAL 302 38 170 0 62 136 708

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

179

Table A6.12

Enterprise Budget for l Hectare of Intensive Local Maize/Beans/Pigeon Peas

 

Quantity Price
(MK)
Output (kg)
Maize 2,049.00 0.330
Beans 151.00 0.650
P/Peas 72.00 0.400
Gross value
Variable Costs
Seed (kg)
Maize 25.00 0.330
Beans 40.00 0.650
P/Peas 6.00 0.400
Fertilizer (kg) 160.00 1.000
Chemicals 1.00 13.000
Marketing (p/h) 2,272.00 0.025
Maint. of Farm Equip. 0.33 19.850
Total Variable Costs
Gross Margin

Labor Input (Median) Person-hours
Month L/Prep Plant Weed Irrig Fert Harv

January 85 34
February

March

April 94
May 94
June

July

August 15 1

September 151

October 19

November 19

December 85 34

TOTAL 302 38 170 0 68 188

Total
1 19

151
151
l9
19
119

766

Value
(MK)

676.17
98.15
28.80

803.12

8.25
26.00
2.40
160.00
13.00
56.80
6.55

260.00

543.12

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

180

Table A6.13
Enterprise Budget for 1 Hectare of Hybrid Maize/Beans/Pigeon Peas

 

 

Quantity Price Value
(MK) (MK)
Output (kg)
Maize 2,999.00 0.330 989.67
Beans 402.00 0.650 261.30
P/Peas 72.00 0.400 28.80
Gross value 1,279.77
Variable Costs
Seed (kg)
Maize 25.00 0.330 8.25
Beans 40.00 0.650 26.00
P/Peas 6.00 0.400 2.40
Fertilizer (kg) 370.00 1.000 370.00
Chemicals 1.00 13.000 13.00
Marketing 3,401 .00 0.025 85 .03
Maint. of Farm Equip. 0.33 19.850 6.55
Total Variable Costs 498.30
Gross Margin 781.49

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 84 169
February 0
March 0
April 140 140
May 140 140
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 84 .169
TOTAL 302 38 170 0 168 280 958

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

181

Table A6.14

Enterprise Budget for l Hectare of Current Fertilized Local Maize/Beans/Groundnuts

 

Quantity Price
(MK)
Output (kg)
Maize 1,212.00 0.330
Beans 92.00 0.650
Groundnuts 186.00 1. 120
Gross value
Variable Costs
Seed
Maize 25.00 0.330
Beans 40.00 0.650
Groundnuts 40.00 1. 120
Fertilizer 115.00 1.000
Marketing 1 ,490.00 0.030
Maint. of Farm Equip. 0.33 19.850
Total Variable Costs
Gross Margin

Labor Input (Median) Person-hours
Month L/Prep Plant Weed Irrig Fert Harv

January 85 29
February
March
April 59
May 59
June
July
August 151
September 151
October 19
‘ November 19
December 85 29

TOTAL 302 38 170 0 58 1 18

Total

114
0

0
59
59
O

O
151
151
19
19
114

686

Value
(MK)

399.17
59.80
208.32

664.29

8.25
26.00
44.80

115.00
37.25
6.55

237.85

426.44

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

182

Table A6.15
Enterprise Budget for 1 Hectare of Intensive Local Maize/Beans/Groundnuts

 

 

Quantity Price Value
(MK) (MK)
Output (kg)
Maize 2,049.00 0.330 676.17
Beans 151.00 0.650 98.15
Groundnuts 186.00 1 . 120 208 . 32
Gross value 982.64
Variable Costs
Seed (kg)
Maize 25.00 0.330 8.25
Beans 40.00 0.650 26.00
Groundnuts 40.00 1 . 120 44.80
Fertilizer (kg) 160.00 1.000 160.00
Marketing (p/h) 2,386.00 0.025 59.65
Maint. of Farm Eqiup. 0.33 19.850 6.55
Total Variable Costs 305.25
Gross Margin 665.46

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 85 40 125
February 0
March 0
April 95 59
May 95 59
June 0
July 0
August 151 151
September 151 151
October 19 19
November 19 19
December 85 40 125
TOTAL 302 38 170 0 80 190 780

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

183

Table A6.16
Enterprise Budget for 1 Hectare of Rice Irrigated Using Gravity/Furrows

 

 

Quantity Price Value

(MK) (MK)

Output (kg) ‘ 3,000.00 0.370 1,110.00
Variable Costs

Seed (kg) 75.00 0.500 37.50

Fertilizer (kg) 230.00 1.000 230.00

Chemicals 1.00 12.000 12.00

Marketing (p/h) 3,300.00 0.025 75.00

Maint. of Farm Equip. 0.33 19.850 6.55

Maint. of Irrig. Equip. 50.00

Inv. in Irrig. Equip. 259.00

Total Variable Costs 670.05

Gross Margin 439.95

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Inig Fert Harv Total
January

February

March

April 151 151
May 151 324 150 625
June 85 150 84 319
July 85 150 84 319
August 298 298
September

October

November

December

TOTAL 302 324 170 450 168 298 1,712

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

184

Table A6.17
Enterprise Budget for 1 Hectare of Rice Irrigated Using Pinup/Sprinklers

 

 

Quantity Price Value

(MK) (MK)

Output (kg) 3,000.00 0.370 1,110.00
Variable Costs

Seed (kg) 75.00 0.500 37.50

Fertilizer (kg) 230.00 1.000 230.00

Chemicals 1.00 12.000 12.00

Marketing (p/h) 3,300.00 0.025 75.00

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip 450.00

Inv. in Irrig. Equip. 2,332.00

Total Variable Costs 3,701.05

Gross Margin -2,591.05

Labor Input (Median) Person-hours

Month L/Prep Plant Weed Irrig Fert Harv Total

January
February
March

April 151 151
May 151 324 150 625
June 85 150 319
July 85 150 319
August 2
September

October

November

December

 

£23?

298

TOTAL 302 324 170 450 168 298 1,712

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

185

Table A6.18
Enterprise Budget for 1 Hectare of Wheat Irrigated Using Gravity/Furrows

 

 

Quantity Price Value

(MK) (MK)

Output (kg) 3,500.00 0.550 1,925.00
Variable Costs

Seed (kg) 100.00 0.550 55.00

Fertilizer (kg) 300.00 1.000 300.00

Marketing (p/h) 3,500.00 0.025 87.50

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 0.00

Maint. of Irrig. Equip. 50.00

Inv. in Irrig. Equip. 259.00

Total Variable Costs 758.05

Gross Margin -l,166.95

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January

February

March

April 151 151
May 151 52 150 353
June 85 150 84 319
July 85 150 84 319
August 347 347
September

October

November

December

TOTAL 302 52 170 450 168 347 1,489

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

186

Table A6.19
Enterprise Budget for l Hectare of Wheat Irrigated Using Pinup/Sprinklers

 

 

Quantity Price Value

Output (kg) 3,500.00 0.550 1,925.00
Variable Costs

Seed (kg) 100.00 0.550 55.00

Fertilizer (kg) 300.00 1.000 300.00

Marketing (p/h) 3,500.00 0.025 87 .50

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 450.00

Inv. in Irrig. Equip. 2,332.00

Total Variable Costs 3,789.05

Gross Margin -1,864.05

Labor Input (Median) Person-hours

 

Month LIPrep Plant Weed Irrig Fert Harv Total
January

February

March

April 151 151
May 151 52 150 353
June 85 150 84 292
July 85 150 84 292
August 347 347
September

October

November

December

TOTAL 302 52 170 450 168 347 1,489

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

187

Table A6.20
Enterprise Budget for l Hectare of Current Relay Rainfed Field Peas

 

Quantity Price Value

Output (kg) 195.00 1.000 195.00
Variable Costs

Seed (kg) 75.00 0.400 30.00

Fertilizer (kg) 0.00 0.00 0.00

Marketing 195.00 0.025 4.88

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 41.43

Gross Margin 153.57

 

Labor Input (Median) Person-hours

 

Month L/Prep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 58 19 77
April 58 19 48 125
May 48 48
June 0
July 28 28
August 28 28
September 0
October 0
November 0
December 0
TOTAL 116 38 96 0 0 56 306

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

188

Table A6.21
Enterprise Budget for l Hectare of Current Dimba Field Peas

 

Quantity Price

(MK)

Output (kg) 670.00 1.000
Variable Costs

Seed (kg) 75.00 0.400

Fertilizer (kg) 0.00 0.000

Marketing 670.00 0.025

Maint. of Farm Equip. 0.33 19.850

Total Variable Costs

Gross Margin

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv

January

February

March 5 15

April 515 38 292

May 292

June 292 120
July 120
August

September

October

November

December

8%

TOTAL 1 ,018 38 128 876 0 240

Total

515
845
356
476
120

OOOOO

2,312

Value
(MK)

670.00
30.00
0.00
16.75
6.55
53.30

616.70

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

189

Table A6.22
Enterprise Budget for l Hectare of Sugarcane

 

 

Quantity Price Value

(MK) (MK)

Value of Output 8,546.00 0.500 4,173.00
Variable Costs*

Marketing (p/h) 8,546.00 0.025 213.65

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 220.20

Gross Margin 4,052.80

Labor Input (Median) Person-hours

Month

January

February
March

April

May

June

July
August
September
October
November
December

TOTAL

LIPrep Plant Weed Irrig Fert Harv Total

274
274
57
57
154

548 114 154

0

OOOOOOO

274

120 394
120 177
57

154

240 1,051

 

* Cuttings are used for propagation and the cost is assumed to be zero.
Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

190

Table A6.23

Enterprise Budget for l Hectare of Current N on-Fertilized Cabbage

 

Quantity Price Value

(MK) (MK)

Output (kg) 5,650.00 0.500 2,825

Variable Costs

Seed (tin) 1.00 20.000 20.00

Fertilizer (kg) 0.00 0.000 0.00

Marketing (p/h) 5,650.00 0.025 141.25

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 167.80

Gross Margin 2,657.20

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 515 515
May 515 124 705 1 ,344
June 124 64 705 893
July 64 705 769
August 705 705
September 240 240
October 0
November 57
December 0
TOTAL 1 ,018 248 128 2,820 0 240 4,466

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

191

Table A6.24

Enterprise Budget for l Hectare of Intensive Cabbage

 

Quantity
Output 22,000.00
Variable Costs
Seed (tin) 1.00
Fertilizer (kg) 80.00
Chemicals (pkt) 1.00
Marketing (p/h) 22,000.00
Maint. of Farm Equip. 0.33
Total Variable Costs
Gross Margin

Labor Input (Median) Person-hours

Price Value
(MK) (MK)
0.500 2,825
20.000 20.00
0.000 80.00
13.000 13.00
0.025 550.00
19.850 6.55
669.55

2,155.45

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 515 515
May 515 124 705 25 1,369
June 124 64 705 25 918
July 64 705 25 799
August 705 25 730
September 240 240
October 0
November 0
December 0
TOTAL 1,018 148 128 2,820 100 240 4,566

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

192

Table A6.25
Enterprise Budget for 1 Hectare of Cabbage Irrigated Using Purnp/Sprinklers-Seasons 2

 

 

Quantity Price Value

Output (kg) 33,481.00 0.500 16,740.50
Variable Costs

Seed (tin) 1.00 20.000 20.00

Fertilizer (kg) 80.00 1.000 80.00

Chemicals (pkt) 1.00 13.000 13.00

Marketing (p/h) 33,841.00 0.025 846.03

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 300.00

Inv. in Irrig. Equip. 1,555.00

Total Variable Costs 3,378.55

Gross Margin 13,361.45

Labor Input (Median) Person-hours

Month L/Prep Plant Weed Irrig Fert Harv Total

January

February

March

April

May 302 248 150 25 725
June 150 25 239
July 150 25 239
August 150 120 270
September

October

November

December

 

ii

TOTAL 302 248 128 600 75 120 1,473

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

193

 

 

 

 

Table A6.26
Enterprise Budget for l Hectare of Cabbage Irrigated Using Pump/Sprinklers-
Seasons land3

Quantity Price Value
Output (kg) 5,069.00 0.500 2,534.50

Variable Costs
Seed (tin) 1.00 20.000 20.00
Fertilizer (kg) 80.00 1.000 80.00
Chemicals (pkt) 1.00 13.000 13.00
Marketing (p/h) 5,069.00 0.025 126.73
Maint. of Farm Equip. 0.33 19.850 6.55
Fuel and Oil 558.00
Maint. of Irrig. Equip. 300.00
Inv. in Irrig. Equip. 1,555.00
Total Variable Costs 2,659.28
Gross Margin -124.78

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total

 

January
February
March

April

May 302 248 150 25 725
June 150 25 239
July 150 25 239
August 150 25 329 504
September

October

November

December

3%

TOTAL 302 248 128 600 100 329 1,707

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

194

Table A6.27
Enterprise Budget for l Hectare of Current Non-Fertilized Tomato

 

 

Quantity Price Value

(MK) (MK)

Output 6,240.00 1.000 6,240.00
Variable Costs

Seed (tin) 1.00 20.000 20.00

Chemicals (pkt) 1.00 13.000 13.00

Marketing (p/h) 6,240.00 0.025 156.00

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 182.55

Gross Margin 6,057.45

Labor Input (Median) Person-hours

 

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 5 15 5 15
May 5 15 658 25 1 ,490
June 64 658 25 747
July 292 64 658 25 722
August 120 120
September 120 120
October 0
November 0
December 0
TOTAL 1 ,030 292 128 l ,974 75 120 3 ,714

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

195

Table A6.28

Enterprise Budget for l Hectare of Intensive Tomato

 

Quantity Price Value

(MK) (MK)

Output (kg) 11,083.00 1.000 11,083.00

Variable Costs

Seed (tin) 1.00 20.000 20.00

Fertilizer (kg) 15.00 0.000 15.00

Chemicals (pkt) 1.00 12.800 12.80

Marketing (p/h) 11,803.00 0.025 295.08

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 348.63

Gross Margin 10,734.37

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 515 515
May 515 658 25 1,490
June 64 658 25 747
July 292 64 658 25 722
August 120 120
September 0
October 0
November 0
December 0
TOTAL 1 ,030 292 128 1,974 75 120 3 ,714

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

196

Table A6.29

Enterprise Budget for l Hectare of Irrigated Tomato Using PrunplSprinklers-Season 2

 

Outpuflkg)

Costs
Seed (tin)
Fertilizer (kg)
Chemicals (pkt)
Marketing (p/h)

Maint. of Farm Equip.

Fuel and Oil

Maint. of Irrig. Equip.

Inv. in Irrig. Equip.
Total Costs

Gross Margin

Labor Input (Median) Person-hours

Month

January

February
March

April

May

June

July
August
September
October
November
December

302

TOTAL 302

292

292

Quantity

10,402.00

1.00
15.00
1.00
10,402.00
0.33

150

64 150
64 150
128 450

25
25

50

Price
(MK)

1.000

20.000
0.000
12.800
0.025
19.850

120

Value
(MK)

10,402.00

20.00
15.00
12.80
260.05
6.55
558.00
300.00
1,555.00

2,727.40

7,674.60

LIPrep Hant Weed Irrig Fert Harv Total

COCO

239
239
120

OCCO

120 1,367

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

197

 

Table A6.30
Enterprise Budget for l Hectare of Irrigated Tomato Using Pump/Sprinklers-
Seasons 1 and 3

Quantity Price Value

(MK) (MK)

Output (kg) 2,471.00 1.000 2,471.00

Costs

Seed (tin) 1.00 20.000 20.00

Fertilizer (kg) 15.00 0.000 15.00

Chemicals (pkt) 1.00 12.800 12.80

Marketing (p/h) 2,471.00 0.025 61.78

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 300.00

Inv. in Irrig. Equip. 1,555.00

Total Costs 2,529.13

Gross Margin -58.13

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 302 292 150 769
June 64 150 25 239
July 64 150 25 239
August 120 120
September 0
October 0
November 0
December 0
TOTAL 302 292 128 450 50 120 1 ,367

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

198

Table A6.31
Enterprise Budget for l Hectare of Current Chinese Cabbage

 

Quantity Price Value

(MK) (MK)

Output (kg) 2,004.00 0.500 1,002.00
Variable Costs

Seed (tin) 1.00 20.00 20.00

Marketing (p/h) 2,004.00 0.025 50.10

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 76.65

Gross Margin 925.35

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 515 515
May 515 292 489 1 .296
June 64 489 553
July 64 489 120 673
August 0
September 0
October 0
November 0
December 0
TOTAL 1 ,030 292 128 1 ,467 50 120 120

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

199

Table A6.32

Enterprise Budget for 1 Hectare of Intensive Chinese Cabbage

 

Quantity

Output (kg) 23,812.00
Variable Costs

Seed (tin) 1.00

Fertilizer (kg) 80.00

Chemicals (pkt) 1.00

Marketing (p/h) 23,812.00

Maint. of Farm Equip. 0.33

Total Variable Costs

Gross Margin

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig F ert

January
February
March

April

May

June

July
August
September
October
November
December

515
515 489
489

489

25
25

292

3%

TOTAL 1,030 292 128 1,467 50

Price Value
0.500 11,906.00
20.00 20.00
1.000 80.00
13.000 13.00
0.025 595.30
19.850 6.55
714.85
11,191.15

Harv Total

0

0

0

515

1,321

120 578

120 673

0

0

0

0

0

120 3,087

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

200

 

Table A6.33
Enterprise Budget for l Hectare of Irrigated Chinese Cabbage Using Pump/Sprinklers-
Season 2
Quantity Price Value
(MK) (MK)
Out!!!It (kg) 23,812.00 0.500 11,906.00
Costs
Seed (tin) 1.00 20.000 20.00
Fertilizer (kg) 80.00 0.000 80.00
Chemicals (pkt) 1.00 13.000 13.00
Marketing (p/h) 23,812.00 0.025 595.30
Maint. of Farm Equip. 0.33 19.850 6.55
Fuel and Oil 558.00
Maint. of Irrig. Equip. 300.00
Inv. in Irrig. Equip. 1,555.00
Total Costs 3,127.13
Gross Margin 8,778.15
Labor Input (Median) Person-hours
Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 302 324 150 801
June 64 150 25 239
July 64 150 25 120 359
August 120 120
September
October 0
November 0
December 0
TOTAL 302 324 128 450 50 240 1 ,494

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

201

 

Table A6.34
Enterprise Budget for 1 Hectare of Irrigated Chinese Cabbage Using Pump/Sprinklers-
Seasons 1 and 3

Quantity Price Value

(MK) (MK)

Output (kg) 65199.00 0.500 3,399.00

Costs

Seed (tin) 1.00 20.000 20.00

Fertilizer (kg) 80.00 0.000 80.00

Chemicals 1.00 13.000 13.00

Marketing (p/h) 6,799.00 0.025 170.00

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 300.00

Inv. in Irrig. Equip. 1,555.00

Total Costs 2,702.53

Gross Margin 696.97

Labor Input (Median) Person-hours

Month L/Prep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 302 324 150 801
June 64 150 25 239
July 64 150 25 120 334
August 120 120
September 0
October 0
November 0
December 0
TOTAL 302 324 128 450 50 240 1 ,494

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

202

Table A6.35

Enterprise Budget for 1 Hectare of N on-fertilized Dimba Maize

 

Quantity Price Value

(MK) (MK)

Output (kg) 2,614.00 0.500 2,614.00

Variable Costs

Seed (kg) 25.00 0.330 8.25

Marketing 2,614.00 0.025 65.35

Maint. of Farm Equip. 0.33 19.850 6.55

Total Variable Costs 80.15

Gross Margin 2,533.85

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 0
June 0
July 515 515
August 515 38 85 608 1,246
September 85 608 693
October 0
November 120 120
December 0
TOTAL 1,030 38 128 1,218 50 120 2,474

 

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

203

Table A6.36

Enterprise Budget for 1 Hectare of Current Non-fertilized Dimba Maize/Pumpkin Leaves

 

Quantity Price Value
(MK) (MK)
Output (kg)
Maize 1,961.00 1.000 1,961.00
Pumpkin Leaves 486.00 0.500 243.00
Gross value 2,204.00
Variable Costs
Seed (kg)
Maize (kg) 25.00 0.330 8.25
P/Lvs (cup) 1.00 1.000 1.00
Marketing 2,447.00 0.025 61.18
Maint. of Farm Equip. 0.33 19.850 6.55
Total Variable Costs 76.98
Gross Margin 2,127.02
Labor Input (Median) Person-hours
Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 5
May 5
June 5
July 515 515
August 515 38 85 608 1 ,246
September 85 608 693
October 0
November 120 120
December 0
TOTAL 1,030 38 128 1,216 50

240 2,574

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

204

 

Labor Input (Median) Person-hours

Month LIPrep Plant Wwd Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 302 38 150 25 441
June 49 150 25 237
July 49 150 160 378
August 160 160
September 0
October 0
November 0
December 0
TOTAL 302 38 128 450 50 320 1 , 198

Table A6.37
Enterprise Budget for 1 Hectare of Irrigated Green Beans Using Pump/Sprinklers-
Season 2

Quantity Price Value
Output 4,716.00 1.000 4,716.00

Variable Costs
Seed (kg) 40.00 1.000 40.00
Fertilizer (kg) 200.00 1.000 200.00
Chemicals 1.00 13.000 13.00
Marketing 4,716.00 0.025 117.90
Maint. of farm lmpl. 0.33 19.850 6.55
Fuel and Oil 558.00
Maint. of Irrig. Equip. 300.00
Inv. in Irrig Equip. 1,555.00
Total Costs 2,790.45
Gross Margin 1,925.55

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

205

 

Table A6.38
Enterprise Budget for l Hectare of Irrigated Green Beans Using Pump/Sprinklers-
Seasons 1 and 3

Quantity Price Value

(MK) (MK)

Output (kg) 3,014.00 1.000 3,014.00

Variable Costs

Seed (kg) 40.00 1.000 40.00

Fertilizer (kg) 200.00 1.000 200.00

Chemicals 1.00 13.000 13.00

Marketing (p/h) 3,014.00 0.025 90.42

Maint. of Farm Equip. 0.33 19.850 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 300.00

Inv. in Irrig. Equip. 1,555.00

Total Costs 2,747.90

Gross Margin 266.10

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 0
May 302 38 150 25 490
June 64 150 25 239
July 64 150 120 359
August 120 120
September 0
October 0
November 0
December 0
TOTAL 302 38 128 450 50 240 1 ,208

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

 

206

Table A6.39
Enterprise Budget for l Hectare of Irrigated Onions Using Pump/Sprinklers-Season 2

 

Quantity Price Value

(MK) (MK)

Output (kg) 15,888.00 1.00 15,888.00
Variable Costs

Seed (tin) 1.00 20.00 20.00

Fertilizer (kg) 79.00 1.00 79.00

Chemicals 1.00 13.00 13.00

Marketing (p/h) 15,888.00 0.03 397.20

Maint. of Farm Equip. 0.33 19.85 6.55

Fuel and Oil 558.00

Maint. of Irrig. Equip. 300.00

Inv. in Irrig. Equip. 1,555.00

Total Costs 2,928.75

Gross Margin 12,959.25

Labor Input (Median) Person-hours

Month LIPrep Plant Weed Irrig Fert Harv Total
January 0
February 0
March 0
April 151 151
May 151 324 150 25 625
June 64 150 25 239
July 64 150 239
August 240 240
September 0
October 0
November 0
December 0

TOTAL 1,030 38 128 50 240 1,494

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

207

Table A6.40

Enterprise Budget for l Hectare of Prunpkin Leaves
Irrigated Using Prunp/Sprinklers-All Seasons

 

Outputfligl

Variable Costs
Seed (cup)
Fertilizer (kg)
Marketing (p/h)
Maint. of Farm Equip.
Fuel and Oil
Maint. of Irrig. Equip.
Inv. in Irrig. Equip.

Total Costs

Gross Margin

Labor Input (Median) Person-hours

Month

January
February
March

April

May

June

July
August
September
October
November
December

302 19

TOTAL 1 ,030 19

3%

128

Quantity
6,626.00

1.00
79.00
6,626.00
0.33

LIPrep Plant Weed Irrig

150
150
150

450

Fert

25
25

50

Price Value
(MK) (MK)
0.500 3,313.00
20.000 20.00
1.000 79.00
0.025 165.65
19.850 6.55
558.00
300.00
1,555.00
2,684.20
628.80

Harv Total

0

0

O

O

585

239

120 334

120 120

0

0

0

0

240 1,278

 

Source: Survey Data 1993/Guide to Agricultural Production in Malawi 1992

 

APPENDIX B

208

THE ROLE OF DIMBA LAND AND SMALL SCALE
IRRIGATION IN IMPROVING HOUSEHOLD FOOD
SECURITY IN SOUTHERN MALAWI

Farmer Survey

 

Name of Enumerator

 

Date of Interview

 

 

I. Identification

Household Code

 

Section Code

 

EPA Code

District Code

 

Name of Respondent

 

209

Name of Enumerator Name of Respondent

 

Date of Interview

 

 

11. Labor Availability

Ql. How many people are in this household?
Q2. Which of these members are available to work on your parcels of land during the
year? (Ask for days in a week, hours in a day.)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Person name
Sex
Age
Relation to Head
September HID
DIW
October HID
DIW
November HID
DIW
December HID
DIW
January HID ll
DIW
February HID
DIW
March HID
DIW
April HID M
DIW
May HID
DIW
June HID ll
W ll

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     

:l I ours per a y - 1 ea
1) = Days per wee Male = 1 0—10 = 1 Head = 0
Female = 2 >10-55 = 2 Wife = 1
>55 = 3 Relative/child = 2

Other = 3

210

III. Soda-Economic Characteristics of Household Heads

Q3. Marital Status
i I = Married 3 = Divorced
' 2 = Widowed 4 = Separated 5 = Single

Q4 Household Type
I = male head 2= female head

Q5. Occupation of Head

1 I = Farming 2 = Employment outside agriculture
1 3 = Agriculture laborer

3 4 = Own business 5 = Unemployed 6 = Other

I

' Q6. Education of Head
1 = No Schooling 2 = Sth-III 3 = Std. IV—VII 4 = Higher
5 = Unknown

 

 

 

 

 

IV. Patterns of Land Use

 

(Put number In code box)

 

Q8(a). Are any of these dimba gardens?
i (Put number in code box. N0 = 0.)

 

Q8(b). Are any of them potentially irrigable? (i.e. within 200 meters of a reliable
water source. N0 = 0.)

 

 

Q9. How did you acquire the dimba land?
I = Inherited from parents
2 = Given by chief
3 = Bought it @rice )

Q10. What type of land is it? (Enumerator to verify.)
I = Close to water source, high water table, well drained
2 = Close to water source, low water table, dries out in dry season
3 = Swampy, poor drainage
4 = Other

Q11(a). Does it flood during the rain season?
Yes = I N0 = 2

Q11(b). Is it waterlogged between April and June?
Yes = I N0 = 2

 

 

 

 

 

 

   

211

MAP

(Draw a map indicating the position of the parcels in relation to household, and for each
parcel of land get details as indicated below to help you fill out tables A, B & C.)

 

 

 

 

 

 

 

 

 

 

 

 

 

flamed; EcheJ—Z

Land type Acres Land Type Acres
Crop Mixture Crop Mixture

(Description of planting) (Description of planting)

No. of Parcel sheets to complete__ N o. of Parcel sheets to complete
Eur—“fl Ear—091.1

Land type Acres Land Type Acres
Crop Mixture Crop Mixture

(Description of planting) (Description of planting)

 

 

 

 

No. of Parcel sheets to complete No. of Parcel sheets to complete

212

Q12(a). Labor Use (1991/1992)

Parcel No.

 

Type ofLand

 

Enterprise

 

Input/Activity

Family Labor

Acres for Enterprise

Crop Mixture

 

 

 

Hired Labor

Months

 

Land Prep

days

 

h/d

 

Pric

 

Peop

 

Planting

days

 

h/d

 

 

Peop

 

Weeding

days

 

h/d

 

Pric

 

Peop

 

Fertilizer

days

 

h/d

 

 

Peop

 

Irrigation

days

 

h/d

 

 

Peop

 

Harvest

days

 

h/d

 

 

Peop

 

Other

days

 

h/d

 

Pric

 

 

 

 

 

213

Q12(b). Input Use (1991/1992)

 

 

 

 

 

 

 

 

Parcel No. Acres for Enterprise only
Type of Land Crop Mixture
Enterprise Months
m \
Price/Unit

 

 

 

 

 

 

 

 

 

 

 

 

Q12(c). Harvest and Sales

 

 

 

 

Parcel No. Acres for Enterprise only
Type of Land Crop Mixture
Enterprise Months

 

 

r___¥fi._.A . _,.__ . 77-. fi. . V- v_ _‘...

Crop—Com; Year Total T“ ‘75“1‘ ‘ m-L______ —— 7
Unit Quantity Unit Quantity Price/Unit l

l

l

l

!

l

1

4

 

1992
1991
1990
1989
1992 ~
1991
1990

1989

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

214

r__ v i ,7 ._ i "in L, ,7 ,_ 7

f Q13. What are the major constraint to producing more of dimba crops? I

I = Water shortage (because the dimba garden normally dries out fast, not I

due to drought) )

l 2 = Lack of fertilizer l

J 3 = Inadequate/unreliable rainfall l
4 = Pests and predators

5 = Other (specify) ’

Q14. What are the major obstacles to producing more of rainfed crops?
2 == Lack of fertilizer
3 = Inadequate/unreliable rainfall
4= Pests and predators
5 = Other (specify)

 

 

 

 

,_.—_.—_.. — ._ _. __ .______.

 

V. Non-Farm Income Generating Activities 5 P

I Q15 What agricultural activity did you engage in between April and September?
I I = Cultivate a second crap on rainfed land

2 = Cultivate dimba land

3 = Other (specify)

Q.16 Are there some months in the calendar year when you do little or no
agricultural activities?
N/A = 999 (Put in code box; otherwise tick months)
|Jan|FeblMarlAprlMaleunlJul|AuglSeplOct|N0v|Dec|

 

 

_—.:——’_—.;_-~.___. _. —....__l

 

Q17(a). Do you use the time to do some income generating activities?
N/A (no flee months) = 999
N0 = 0
Farm IGAS = 1

Other (specify) =

 

 

 

215

Name of Enumerator Name of Respondent

 

Date of Interview

 

17(b). What is the income per week/month from this activity?
MA = 999
Record figure per month
(Convert per day/week to per month)

 

. What are the prerequisites to engage in the activity?
N/A = 999
(Open ended) If it is a trade list training or apprenticeship required or
amount of investment/type of equipment required.

 

 

 

 

 

; Q17(d). How many days a month do you engage in the activity? (Put No. of Days)
N/A = 999

216

VI. Willingness to Use and Invest in Irrigation

 

j Q18(b). Suppose you were to invest in irrigation technology of

' cost for a quarter of an acre (both capital and interest) and a
repayment period of two years. Your installments would be MK200 every
three months.

 

Would you be willing to make the investment?

 

What if your net income from using the one quarter of an acre were:

MK200.00. Would you be willing to make the investment?

 

MK500.00?
Q18(c). If ’no’, why? (Open ended)

 

 

 

 

 

 

217

Q19. If you invest in irrigation which crops would you want to grow on your dimba? When
and why? (Give names of two most important crops.)

' Season 1. October — March
1 Crop (mixture)
I Why

 

 

ll Season 2. April - June

Crop (mixture)
Why

 

 

 

Season 3. July - September

I Crop (mixture)
‘ Why
|__, __

 

 

Codes for wh

Water will be available
For food

For cash

Suitable for soils

Other (specify)

mature—1
II II II II II

 

218

Q20. [If yes to Q8(b)] If you invested in irrigation what crops would you like to grow on
this piece of land that can be irrigated? (Note that the land can then be cropped from
April to September.)

I Season]. OctOber- March
I
I

Crop (mixture)
Why

 

I

I

I

I

Season 2. April - June I
. ‘ I
Crop (mixture) I
I

Why

 

Crop (mixture) I
Why ’

 

I
I
I
I
I
I
Season 3. July - September I
I
I

Q21(a). Is your dimba parcel of land contiguous with other farmers’ dimba
gardens?
1 = yes 2 = No

 

Q21(b). Would you be willing to participate in a communal irrigation project?
I = Yes 2 = No

 

Q21(c) Do you foresee any problems in setting up and organizing communal
irrigation projects? (Open ended) ‘

 

 

 

 

219
VII. Use of Extension and Credit Services

Q22(a). In the last 3 years have you received credit from any source?

(If no crdt ten 999 all through) _

 

Condition

 

Interest

 

1991/92

 

 

 

 

 

 

 

. 1989/90

Q23(a). In the past 3 years have you been in contact with any agriculture/irrigation extension
agent?

frequnet

 

1991/92
1990/91
1989/90

 

 

 

 

 

 

Q23(b). If no, why not? (Open ended)

 

220

VIII. Food Production Strategies and Security

0 = Produce some food/grow cash crops
1 = Produce all food requirements

2 = Do ganyu labor (where
3 = Member(s) of family permanently employed
4 = IGAs or business (specify
5 = Other (specify)

Q24(b). If you rely on agricultural production (codes 0 and 1), what do you
do when rain fails to come in good time/amounts?

N IA = 999, i.e. does not rely on agricultural production
1 = Rely on dimba productions
2 = Trading in an agricultural commodity
3 = Practicing a trade (specify)
4 = Other (specify)
5 = Selling labor (specify)

 

 

 

 

. How many months did your food production last in the past three years?
(Put the no. of months in the code boxes.)

M_Ot1_tlls_las_t¢.cl

 

 

221

GARDEN MEASUREMENT

Name of Enumerator ................ Date ................
Name of Assistant ................. Section Code ...........
Name of Respondent ................ Household Code .........

 

 

 

 

 

 

 

 

TOTAL AREA ha acres
Perimeter meters
Plot 1: Area ha acres Plot 2: Area ha acres
Perimeter meters Perimeter meters
N/S error N/S error
EW error EW error
% gap error % gap error
Plot 3: Area ha acres
Perimeter meters
NI S error
EW error

% gap error

Garden
Stations
From - To

Forward

Length
of Side

Backward
Bearing

Plot
No.

222

Plot
Stations
From - To

Forward
Bearing

Length
of Side

Cumulative
Length of
Sides

Backward
Bearing

 

A-B

 

B-C

 

C-D

 

D-E

 

E-F

 

F-G

 

G-H

 

H-I

 

 

 

 

 

 

N-O

 

O-P

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

BIBLIOGRAPHY

BIBLIOGRAPHY

Abemethy, C. L. Methodology for Studies of Irrigation Water Management, in
African Regional Symposium on Smallholder Irrigation, M. J. Blackie ed.,
Harare: University of Zimbambwe, Department of Land Management, 1984,

pp 1—8.

Adams, R. M., D. J. Menkhaus and B. A. Woolery, "Alternative Parameter
Specialization in E-V Analysis: Implication for Farm Level Decision Making",
Western Journal of Agricultural Economics, 5(1980):13-20.

Anderson, J. R., J. L. Dillon and J. B. Hardaker, Agricultural Decision Analysis,
Ames Iowa, Iowa University Press, 1977.

Anderson, J. R. , "Risk Efficiency in the Interpretation of Agricultural Production
Research", Review of Marketing and Agricultural Economics 42(1974): 131-84.

Apland, J. D., B. A. McCarl, and W. L. Miller. "Risk and the Demand For
Supplemental Irrigation: A Case Study in the Com-Belt" , American Journal of
Agricultural Economics, 62(1980):142-145.

Arrow, K. J ., Aspects of the Theory of Risk Bearing, Yrjo Jahnasson, Lectures, Yrjo
Jahnasso in Saatio, Helsinki, 1965.

Arrow, K. J. , "Use of Unbounded Utility Functions in Expected Utility
Maximization-Response", Quarterly Journal of Economics, 88(1974):136-138.

Atwood, J. , "Demonstration of the Use of Lower Partial Moments to Improve Safety—
First Probability Limits " , American Journal of Agricultural Economics,
67(1985):787-793.

Atwood, J ., and J. Segar, "An Application of Safety First Probability Limits in a
Discrete Stochastic Farm Management Programming Model" , Southern
Journal of Agricultural Economics, 21 (1989):65-72.

Atwood, J ., M. J. Watts, G. A. Helmers, and L. J. Held, "Incorporating Safety-First

Constraints in Linear Programming Production Models" , Western Journal of
Agricultural Economics, 13(1988):29—36.

223

224

Atwood, J ., M. J. Watts, and G. Helmers, "Safety First Models Based on Sample
Statistics", Risk Analysis for Agricultural Production F inns: Concepts,
Information Requirements and Policy Issues, Staff Paper No. 85-58
Department of Agricultural; Economics, Michigan State University, East
Lansing, 1985.

Atwood, J., M. J. Watts and G. Helmers, "Chance Constraint Financing as a
Response to Financial Risk", American Journal of Agricultural Economics,
70(1988):79-89.

Babcock, B. A., J. A. Chalfant and R. N. Collender, "Simultaneous Input Demands
And Land Allocation to Agricultural Production Under Uncertainty", Western
Journal of Agricultural Economics, 12(1987):207-215.

Balarin, J. D. , "Fish Farming and Irrigation: Integrated Techniques for Protein
Production" , African Regional Symposium on Smallholder Irrigation, M. J.
Blackie (ed.), Harare: University of Zimbambwe, Department of Land
Management, 1984 pp 151-165.

Barnett, T. , "Small Scale Irrigation in Sub-Saharan Africa: Sparse Lessons, Big
Problems, Any Solutions? " , Public Administration and Development
4(1984):21-47.

Baron, D. P. , "On the Utility Theoretic Foundations of Mean-Variance Analysis",
The Journal of Finance, 5( 1977): 1683-1697.

Barry, P. J., Risk Management in Agriculture, Iowa State University Press, Ames,
1984.

Barry, P. J. and L. J. Robinson, "A Practical Way to Select an Optimum Farm Plan
Under Risk: Comment" , American Journal of Agricultural Economics,
58( 1975): 128-31 .

Barry, P. J. and D. R. William, "A Risk Analysis of Forward contracting with Credit
Constraints", American Journal of Agricultural Economics, 58(1976):62-70.

Baumol, W. 1'., "An Expected Gain-Confidence Limit Criterion for Portfolios
Selection", Management Science, 10(1963): 174-182.

Beneke R. R. and R. Winterboer, Linear Programming Applications to Agriculture,
Iowa, Iowa State University Press, 1973.

Berck, P. and J. M. Hihn, "Using Semi-Variance to Estimate Safety First Rules",
American Journal of Agricultural Economics, 64(1982):298-300.

Binswanger, H. P. , Attitudes Towards Risk: Experimental Measurement in Rural
India, American Journal of Agricultural Economics, 62(1980):395-407.

225

Biswanger, H. P., and D. A. Sillers, "Risk Aversion and Credit Constraints in

Farmers Decision-making: A Reinterpretation" , Journal of Development
Studies, 20(1983).

Boisvert, R. , Choice in Risk Situations: A Comparison and Evaluation of Alternative
Criteria, Staff Paper No. 72-14, Department of Agricultural Economics,
Cornell University, Ithaca, May 1972.

Boisvert, R. , The Role of Alternative Risk Programming Models in Empirical
Research, Staff Paper No.85-85, Department of Agricultural Economics,
Michigan State University, East Lansing, November 1985.

Boisvert, R. N. and Bruce McCarl, Agricultural Risk Modelling Using Mathematical
Programming, Department of Agricultural Economics, Cornell University,
Southern Cooperative Series: Bulletin No. 356, 1990.

Bollamn, R. D. , Off-farm Work by Farmers: An Application of the Kinked Demand
Curve for Labor, Canadian Journal of Agricultural Economics, 27(1979):37-
60.

Brandao, E., B. A. McCarl and G. E. Schuh, "Predicting the Impact of New Crop
Practices Upon Subsistence Farming: A Farm level Analysis in Brazil" ,
Western Journal of Agricultural Economics, 9 (1984):329-341.

Brink, L. and B. A. MacCarl, "The Trade-off Between Expected Return and Risk
Among Corn Belt Farmers" , American Journal of Agricultural Economics,
60(1978):713-41.

Brink, L. and B. A. McCarl "The Adequacy of a Crop Planning Model for
Determining Income Change, and Crop Mix" , Canadian Journal of
Agricultural Economics, 27(1979):13-15.

Buccola, S. T. , "Minimizing Mean Absolute Deviations to Exactly Solve Expected
Utility Problems: Comment" , American Journal of Agricultural Economics,
64(1982) : 789-91 .

Carnap, R. , Logical Foundations of Probability, Second edition, Chicago University
Press, Chicago, 1962.

Coase, R. H., "The Problem of Social Cost", Journal of Law and Economics,
3(1960):1-44.

Casley, D. J. , and D. A. Lury, Data Collection in Developing Countries, Claredon
Press, Oxford, 1981.

Center for Social Research, Report of the Workshop on Food Security and Nutrition,
Zomba, Malawi, 1988.

 

226

Chalfant, J. A., R. N. Collender and S. Subramanian, "The Mean and Variance of
the Mean-Variance Rule" , American Journal of Agricultural Economics,
72(1990):966-974.

Chames, A. and W. W. Cooper, "Chance Constrained Programming", Management
Science, 6(1959):73-79.

Chipande, G. H. R. , "Innovation Adoption Among Female-headed Households " ,
Economic Development and Cultural Change, 18(1987):315-327.

Chipeta, M., Political Process, Civil Society and the State, in Malawi at the
Crossroads: The Post Colonial Political Economy, G. Mhone (ed.), SAPES
Books, Harare, 1992, pp2 31.

Christiansen, R. and J. Kydd, The Political Economy of Agricultural Policy in
Malawi, World Bank MAIDA Paper, Washington, DC, 1985.

Churchman, C. W. , Prediction and Optimal Decisions: Philosophical Issues of a
Science of Value, Prentice Hall, Englewoods Cliffs, 1961.

Cochran, M., W. Lodwick, and L. Robinson, Rankings of Action Choices Under
Uncertainty: A Discussion of Stochatic Dominance Techniques, Staff Paper No.
1982-68, Michigan State University, Department of Agricultural Economics,
East Lansing, 1982.

Cocks, K., "Discrete Stochastic Programming", Management Science, 15(1968):72-
79.

Collender R. N. and J. J. Chalfant, "Alternative Approaches to Decisions Under
Uncertainty Using the Empirical Moment Generating Function", American
Journal of Agricultural Economics, 68(1986):727-731.

Collender, R. N., "Estimation Risk in Farm Planning Under Uncertainty", American
Journal of Agricultural Economics, 71(1989):996-1002.

Connor, J. R., R. J. Freund and M. R. Goodwin, "Risk Programming: An Aid in
Planning Reservoir Systems", American Journal of Agricultural Economics,
54(1972):249-254.

Cornea, G. A. , R. Jolly and F. Stewart, Adjustment with a Human Face, Claredon
Press, Oxford, 1987.

Crawford, E. , Class Notes: Managing Agricultural Production in Developing
Countries, 1992.

Day, R. , "Probability Distribution of Field Yields" Journal of Farm Economics,
47(1964): 713-41.

Debertin, D. L. , Agricultural Production Economics, Macmillan Publishing Company,
New York, 1986.

227

De Janvry, A. , "Optimal Levels of Fertilization Under Risk: The Potential for Corn
and Wheat Fertilization Under Alternative Price Policies in Argentina",
American Journal of Agricultural Economics, 55(1972):1-10.

Dey, J ., Women and Rice in the Gambia: The Impact of Irrigated Rice Development
Projects on Farming Systems’. Ph.D. Thesis, University of Reading (quoted
by Barnett, 1984).

Diallo, M. , Comparative Analysis of Capital and Labor Intensive Rice Perimeters in
the Senegal River Valley, MS Thesis, Michigan State University, 1980.

Dillon, J. L. and J. B. Hardaker, Farm Management Research for Small Farmer
Development, FAO, Agricultural Services Bulletin N o. 41.

Dillon, J. and P. Scandizzo, "Risk Attitudes of Subsistence Farms in Northeast

Brazil: A Sampling Approach" , American Journal of Agricultural Economics,
60(1978):425-435.

Doll, J. P. and F. Orazem, Production Economics: Theory with Applications, New
York, Wiley, 1984.

Eicher, C. K. and Daler, D. C. , Research on Agricultural Development in Sub- .
Saharan Afiica: A Critical Survey, East Lansing, Michigan State University,
International Development Paper N o. 1, 1982.

El-Nazer, T. and B. A., McCarl. "The Choice of Crop Rotation: A Modeling
Approach and Case Study", American Journal of Agricultural Economics,
68(1986):127-136.

Ellis, F. , Peasant Economics, Farm Households and Agrarian Development,
Cambridge University Press, Cambridge, New York, 1988.

Eastern Software Products, LP88 User ’s Manual, Alexandria, Virginia, I987.

Epperson, J. C., and L. F. Ici, "A Regional Analysis of Vegetable Productions with
Changing Demand for Row Crops Using Quadratic Programming", Southern
Journal of Agricultural Economics, 21(1989):87-96.

Faki, H. H. , Economics and Management of Irrigation Projects in Sudan Gezira
Scheme, Ph.D. Dissertation, University of Hohenheim, Hohenheim, 1982.

F AO, Design and Operation of Irrigation Systems for Smallholder Agriculture in
South Asia, Vol. 1 and 2, Rome, 1986(b).

Feder, G. "Farm Size, Risk Aversion and the Adoption of New Technology Under
Uncertainty" Oxford Economics Papers, 32(1980):263-83.

Feder, G. , R. E. Just, and D. Zilberman, "Adoption of Agricultural Innovations In
Developing Countries: A Survey", Economic Development and Cultural
Change, 33(1985):255«298.

' 228

F ishbum, P. C. "Convex Stochastic Dominance with Continuous Distribution
Functions", Journal of Economic Theory, 7(1974):143-158.

Fishburn, P. C. "Mean Risk Analysis with Risk Associated with Below Target
Returns", American Economic Review, 67(1977):116-23.

Fishbum P. C., Non-transitive Measurable Utility, Economics Discussion Paper No.
209, Bell Laboratories, Murray Hill, New Jersey, 1981 (quoted in Machina,
1982).

Fishburn, P. C. , "Foundations of Risk Measurement: Risk as Probable Loss",
Management Science, 30(1984):396—406.

Freund, R. J. , "The Introduction of Risk in a Programming Model", Econometrica 24
(1956):25-64.

Friedman, M. and L. Savage, "The Utility Analysis of Choices Involving Risk",
Journal of Political Economy, 569(1948):279-304.

Griliches, Z. , "Hybrid Corn: An Exploration in the Economics of Technology
Change", Econometrica, 25(1957):501-22.

Hacking, 1., The Emergence of Probability, Cambridge, Massachussetts, Cambridge
University Press, 1975.

Hadar, J. and W. R. Russell, "Rules for Ordering Uncertain Prospects", American
Economic Review, 59(1969):25-34.

Hadar, J. , and W. R. Russell, ”Diversification of Interdependent Prospects", Journal
of Economic Theory, 71(1974):231-40.

Hanoch, G., and H. Levy, "The Efficiency Analysis of Choices Involving Risk",
Review of Economic Studies, 36(1969):335-46.

Hansotia, B. J. , "Stochastic Linear Programming with Recourse: A Tutorial",
Decision Science, ll(1980):151-168.

Harsh, S. B., Larry J. Connor, and Gerald D. Schwab, Managing the Farm Business,
Englewoods, New Jersey: Prentice-Hall, 1981.

Harvey, C. , Successful Macro-Economic Adjustment in Three Developing Countries:
Botswana, Malawi, and Papua New Guinea, Economic Development Institute,
1987.

Hazel], P. B. R. , "A Linear Alternative to Quadratic Programming and Semivariance
Programming for Farm Planning Under Uncertainty" , American Journal of
Agricultural Economics, 53(1971):53-62.

Hazel], P. B. R. and R. D. Norton, Mathematical Programming for Economic
Analysis in Agriculture, MacMillan Publishing Company, 1986.

 

229

Holthausen, D. M. , "A Risk—Retum Model with Risk and Return Measured as
Deviations from A Target Return", American Economic Review, 71(1981):182—
88.

Holtzman, J. S. , "Maddening Myths About Agricultural Marketing in Developing
Countries", Journal of International Food and Agribusiness Marketing,
1(1989).

Huntington Technical Services Limited and Malawi Government, Ministry of
Agriculture, National and Shire Irrigation Project, Final National Report,
Lilongwe, June 1981.

Ibnouf, M. A. , An Economic Analysis of Mechanized Food Production Schemes in
Central Plains of the Sudan, Ph.D. Dissertation, Department of Agricultural
Economics, Michigan State University, East Lansing, 1985.

Islam, T. R. , Alternative Low Lifl Pump Irrigation Policies in Bangladesh: An
Economic and Financial Analysis, Ph.D. Dissertation, Michigan State
University, East Lansing, 1973.

Jaffee, S. R. Mkandawire, and S. Bertoli, The Migrant Smallholders: Tenant and
Labor Participation, Remuneration, and Social Welfare Within Malawi ’s
Expanding Estate Sub-sector, Institute of Development Anthropology, Clark
University, 1991.

Jere, O. , Evaluation of Smallholder Agricultural Technology in Malawi, MS Thesis,
Department of Agricultural Economics, Michigan State University, East
Lansing, 1990.

Johnson, G. , Research Methodology for Economists Philosophy and Practice,
Macmillan Publishing Company, New York, 1986.

Johnson, D. , and M. Boehlje," Minimization Mean Absolute Deviation to Exactly
Solve Expected Utility Problems" , American Journal of Agricultural
Economics, 63(1981):728-729.

Just, R. E., "An Investigation of the Importance of Risk in Farmers Decisions",
American Journal of Agricultural Economics, 56(1974):14-25.

Just, R. E., "Risk Aversion Under Profit Maximization", American Journal of
Agricultural Economics, 57(1975):347-352.

Kahnert, F. and G. Levine, Ground Water Irrigation and the Rural Poor: Options for
Development in the Gangetic Basin, A World Bank Symposium, 1993.

Kaisser, H. M., and J. Apland. "Discrete Stochastic Programming: A Model of
Production and Marketing Decision on a Midwestern Crop Farm" North
Central Journal of Agricultural Economics, 11(1989): 157-170.

 

230

Kalua, B. , Malawi Food Marketing: Private Traders and State Intervention,
Chancellor College, Zomba, Malawi.

Kampman, K. , Risk Analysis of Subirrigation Investment Decisions in the Saginaw
Bay Area of Michigan, MS Thesis, Department of Agricultural Economics,
Michigan State University, East Lansing, 1993.

Kandoole, B. , Structural Adjustment in Malawi: Short Run Gains and Long Run
Losses, Chancellor College, Zomba, 1989.

Kataoka, S., "A Stochastic Programming Model" Econometrica 31(1963):181-96.

King, R. , Operational Techniques for Applied Decision Analysis Under Uncertainty,
Ph.D. Dissertation, Michigan State University, East Lansing, 1979.

Knight, F. H., Risk, Uncertainty and Profit, Boston: Houghton Miffm, 1921.

Kydd, J., "Maize Research in Malawi: Lessons From Failure", Journal of
International Development, 1(1989):112-144.

Kydd, J . , and R. Christiansen, "Structural Change in Malawi Since Independence:
Consequences of a Development Strategy Based on Large-Scale Agriculture"
World Development, 1982.

Kroll, Y. H., H. Levy and H. M. Markowitz, "Mean Variance Versus Expected
Utility Maximization", Journal of Finance, 39(1984):47—62.

Laki, S. L. , Policy Analysis of Sudan ’s Irrigated Subsector: The Case of Gezira
Irrigation Scheme. Ph.D. Dissertation, Department of Agricultural Economics,
Michigan State University, 1992.

Lambert, D. K. and B. A. McCarl, "Sequential Modelling of Wheat Marketing
Strategies", North Central Journal of Agricultural Economics, 11(1989): 105-
115.

Lambert, D. K. and B. A. MacCarl, "Risk Modelling Using Direct Solution of Non-
Linear Approximations of the Utility Function" , American Journal of
Agricultural Economics, 67(1985):846-52.

Lee, J . , D. Brown and S. Lovejoy, "Stochastic Efficiency versus Mean Variance
Criteria", American Journal Agricultural Economics, 67(1985):839-846.

Lele, U. , The Design of Rural Development: Lessons from Africa, Baltimore,
Maryland, John Hopkins University Press, 1975.

Lele, U. , Structural Adjustment, Agricultural Development and the Poor: Lessons
from the Malawian Experience, World Bank MADIA Paper, Washington,
DC. , 1989.

231

Lele, U. , Smallholder and Large Scale Agriculture in Africa: Are There Trade-017‘s
Between Growth and Equity, World Bank MADIA Paper, Washington, DC,
1989.

Lele, U. and S. W. Stone, "Population Pressure, the Environment and Agricultural
Intensification in Sub-Saharan Africa: Variation of the Boserup Hypothesis,
MADIA Discussion Paper No.4, The World, Washington, DC, 1989.

Levy, H. , "The Rationale of Mean-Standard Deviation Analysis: Comment" ,
American Economic Review, 64(1974):434-41.

Levy, H. and Markowitz, H., "Approximating Expected Utility by a Function of
Mean and Variance", American Economic Review, 69(1979):308 -17.

Lilongwe Agricultural Development Division, Minutes of the Irrigation Committee,
February 18th, 1993(a). .

Lilongwe Agricultural Development Division, Report of Field Ofi‘icer (Horticulture) to
Project Officer, Lilongwe East Rural Development Project, February 28th,
1993 (b).

Lipton, M. , "The Theory of the Optimizing Peasant", Journal of Development
Studies, 4(1968):372-51.

MacCrimmon, K. R. and L. Stig, "Utility Theorem: Axioms Versus ’Paradoxes’ "
Expected Utility and the Allais Paradox, M.Allais and O. Hagen (eds.)
Dordrecht, Holland, 1979, pp3333-409.

Machina, M. , "New Theoretical Developments: Alternatives to Expected Utility
Maximization" , Risk Analysis for Agricultural Production Firms: Concepts,
Information Requirements and Policy Issues, Staff Paper No. 85-85,
Department of Agricultural Economics, Michigan State University, East
Lansing, 1985.

Machina, M. , "Expected Utility-Analysis Without the Independence Axiom",
Econometrica, 50(1982):277-323.

Machina, M. , "Choice Under Uncertainty: Problems Solved and Unsolved", Journal
of Economic Perspective 1(1987):121-154.

Maji, C. , and E. Heady, " Intertemporal Allocation of Irrigation Water in the
Mayurakshi Project (India): An Application of Chance Constrained Linear
Programming", Water Resources Research, 14(1978):190-205.

Makhado, J. M. , "A Review of the Factors Affecting the Viability of Smallholder
Irrigation Schemes in Africa" , in African Regional symposium on Smallholder
Irrigation, M. J. Blackie (ed.), Harare: University of Zimbambwe,
Department of Land Management, 1984.

Makeham D. and L. R. Malcolm, Economics of Tropical Farm Management, 1986.

 

 

 

 

 

232

Malawi Government, National Statistical Office, National Sample Survey of
Agriculture, Government Printer: Zomba, 1981.

Malawi Government, MiniStry of Agriculture, Annual Survey of Agriculture,
Lilongwe, 1985.

Malawi Government, Statement of Development Policies, Government Printer:
Zomba, 1979 and 1989.

Malawi Government, Ministry of Agriculture, Lower Shire Valley Irrigation
Development Project-Main Report - Assessment on Present Environment, Land
Tenure and Land Allocation Study, Lilongwe, 1989.

Malawi Government, Food Security and Nutrition Unit, Food Security and Nutrition
Bulletin, July 1990, Vol.2, No.1.

Malawi Government, Ministry of Agriculture, Lilongwe ADD, Food Security and
Nutrition Monitoring Survey, Lilongwe, March 1991.

Malawi Government, Ministry of Agriculture, National Early Warning System for
Food Security, Lilongwe, January 1991.

Malawi Government, Ministry of Agriculture, Study on Self-Help Irrigation
Communities: Description of Sites In Mzuzu and Blantyre ADDs, Lilongwe,
January 1992(a).

Malawi Government, Ministry of Agriculture, Study on Self-Help Irrigation
Communities: The Implementation of Self-Help Schemes, Lilongwe, September
1992(b).

Malawi Government, Ministry of Agriculture, A Guide to Agricultural Production In
Malawi, Lilongwe, 1992(c).

Markowitz, H. M., Mean-Variance Analysis in Portfolio Choice and Capital Markets,
Basil Blackwell, New York, 1987.

Marra, M. C. and G. A. Carlson, "The Role of Farm Size and Resource Constraints
in the Choice Between Risky Technologies", Western Journal of Agricultural
Economics, 12(1987):109-18.

McCamley, F., and J. B. Kliebenstein, "Safety First Models Based on Sample
Statistics: A Discussion", in Risk Analysis for Agricultural Production Finns:
Concepts, Information Requirements and Policy Issues, Staff Paper No. 85-85,
Department of Agricultural Economics, Michigan State University, East
Lansing, 1985.

McCamley, F. and J. B. Kliebenstein, "Describing and Determining the Complete Set
of Target MOTAD Solutions" , American Journal of Agricultural Economics,
69(1987):669-76.

 

233

McCarl, B. A. , and H. Onal, "Linear Approximations Using MOTAD and Separable
Programming: Should It Be Done?" , American Journal of Agricultural
Economics, 71(1989):158-165.

McCarl, B. A. and T. Tice, "Should Quadratic Programming Problems Be
Approximated?", American Journal of Agricultural Economics, 64(1982):585-
589.

McCarl, B. A., "Degeneracy, Duality and Shadow Prices in Linear Programming",
North Central Journal of Agricultural Economics, 25(1977):70-73.

McCarl, B. A. and G. H. Parandvah, "Irrigation Development Versus Hydroelectric
Generation: Can Interruptable Irrigation Play A Role?" , Western Journal of
Agricultural Economics, 13(1988):267-276.

Meyer, J. , "Choice Among Distributions", Journal of Economic Theory,
14(1977):326-336.

Meyer, J. , "Second Degree Stochastic Dominance with Respect to a Function" ,
International Economic Review, 18(1977):477-87.

Meyer, J. , "Two-Moment Decision Models and Expected Utility Maximization",
American Economic Review, 77(1987):421-30.

Meyers, K. K. , Emerging Malawi, Problem of Developing Countries in the 1990s,
Vol II- Country Studies, F. D. Mac Carthy (ed) World Bank, Washington,
DC. , 1990. ’

Mhone, G. , "The Political Economy of Malawi: An Overview" , Malawi at the
Crossroads: The Post Colonial Political Economy, G. Mhone (ed.), SAPES
Books, Harare, 1992.

Mian, M. S. , Problem and Prospects for Power Pump Irrigation in Some Selected
Thanas of Camilla District, Bangladesh, Academy for Rural Development,
Kotbari, Comilla, Bangladesh, 1976.

Ministry of Agriculture, Blantyre Agricultural Development Division, Smallholder
Agricultural Credit Administration, National Credit Package, Price List,
1989/90.

Lilongwe, Malawi, 1990.

Mkandawire, R. S. Jaffee and S. Bertoli, Beyond Dualism: The Changing Face of the
Leasehold Estate Sector in Malawi, Institute for Development Anthropology,
Clark University, 1990.

Msukwa, L. A. H., B. F. Kandoole, and W. Chilowa, Survival Mechanism an
Intervention Strategies of Food Deficit Households, Centre for Social Research,
Zomba, Malawi, 1991.

234

Musser, W., N. J. Ohannesian, and F. J. Benson, "A Safety First Model of Risk
Management for Use in Extension Programs, North Central Journal of
Agricultural Economics, 3(1981):41-46.

Musser, W., V. J ., Alexander, B. V. Teo and D. A. Smittle, "A Mathematical
Programming Model for Vegetable Rotations, Southern Journal of Agricultural
Economics", l7(1985):169-76.

Ndiame, Fadel, A Comparative Analysis of Alternative Irrigation Schemes and the
Objective of Food Security: The Case of The Flueve in Senegal, MS. Thesis,
1985.

Nankumba, J. , Evolution of the Agricultural Policy in Malawi and the National Rural
Development Program: A Historical Review (1891-1980) , Department of
Agricultural Economics, Abertywynth, University of Wales, 1981.

Ngwira, N. , The Efl'ectiveness of Agricultural Extension Service in Reaching Rural
Women with Appropriate And Timely Agricultural Information, FAQ/Center
For Social Research, Zomba, Malawi, May 1987.

Ngwira, N ., Chilowa, W., Asiedu, J ., and Msukwa, L., Income Generating Activities
for Women in Karonga, Kasungu, and Mangochi Districts, Center for Social
Research, Malawi, 1988.

Ngwira, Naomi, Organization and Management of Agricultural Services for Small
Farmers in Malawi, FAQ-Center for Social Research, Zomba, Malawi, 1988.

N gwira, Naomi, A Profit Model for Predicting Maize Self-Suficiency Amongst Rural
Households in Southern Malawi, Mimeo, East Lansing, May 1992.

North, D. , Institutions, Institutional Change and Economic Ped'ormance: The Political

Economy of Institutions and Decisions, Cambridge University Press,
Cambridge, 1990.

Paris, Q. , "Multiple Optimal Solutions in Linear Programming Models", American
Journal of Agricultural Economics, 63(1981):724-729.

Patrick, G. R., P. N. Wilson P. J. Barry, W. G. Boggess and D. L. Young, "Risk
Perceptions and Management Responses: Producer-Generated Hypothesis for
Risk Modelling", Southern Journal of Agricultural Economics, 2(1985):231-
238.

Pingali, P. , Y. Bigot, and H. P. Binswanger, Agricultural Mechanization and the
Evolution of Farming Systems in Sub Saharan Afiica, John Hopkins University
Press, Baltimore, 1987. '

Peters, P. E. and M. G. Herrera, Cash Cropping, Food Security and Nutrition: The
Effects of Agricultural Commercialization Among Smallholder Farmers in
Malawi, Harvard Institute for International Development, Cambridge, MA,
June 1989.

235

Peters, P. E. , Monitoring the Efl’ects of Grain Marketing Liberalization on the
Income, Food Security and Nutrition of Rural Households in Zomba South,
Malawi, Harvard Institute for International Development, Cambridge, MA,
August 1992.

Phillips, L. D. and C. N. Wright, "Cultural Differences in Viewing Uncertainty and
Assessing Probabilities", in Decision Making and Change in Human Afiairs,
H. Jungermann and G. de Zeeuw (eds.), Dordrecht, Holland and Boston,
1977, pp 507-10 (quoted in Machina 1982).

Porter, R. B. , "Semivariance and Stochastic Dominance: A Comparison", American
Economic Review, 64(1974):200—204.

Arrow, K. J. , The Theory of Risk Aversion: Essays in the Theory of Risk Bearing,
Chicago: Markham Publishing Company, 1971.

Pratt, J. W. "Risk Aversion in the Small and in the Large", Econometrica,
32(1964): 122-36.

Priebprom, S. , Employment and Income fiom Farming, Non-farm Enterprises and 0]?-
farm Work on Irrigated and Rainfed Farms in Khon Kaen Thailand, Ph.D.
Thesis, Michigan State University, East Lansing 1982.

Pyle, D. H. and S. J. Tumovsky, "Safety First and Expected Utility Maximization in
Mean-Standard Deviation Portfolio Analysis" Review of Economics and
Statistics 52(1970):75-81.

Rae, A. N. , "Profit Maximization and Imperfect Competition: An Application of

Quadratic Programming to Horticulture" , Journal of Agricultural Economics,
21(1970):133-140.

Reddy, V. K. and J. K. Finders, "Determinants of Off-Farm Labor Force
Participation: Implications for Low Income Farmers", North Central Journal
of Agricultural Economics, 10(1988)91-102.

Rinaldi, A. , Irrigation in Southern Africa: An Annotated Bibliograph)’, Michigan State
University, Department of Agricultural Economics, Working Paper No. 19,
East Lansing, 1985.

Rothschild, M., and Stiglitz, J. E., "Increasing Risk 1: A Definition", Journal of
Economic Theory, 2(1970):225-43.

Roumasset, J. , Rice and Risk: Decision Making Among Low Income Farmers,
Amsterdam: N orthland Publishing Company, 1976.

Roy, A. D., "Safety-First and Holding of Assets", Econometrica 20(1952):43-49.

Russell, W. R. , and T. K. Seo, "Admissible Sets of Utility Functions in Expected
Utility Maximization", Econometrica, 46(1978):181-84

 

236

Samuelson, P. A., "General Proof that Diversification Pays" Journal of Financial
Quantitative Analysis, 2(1967):401-416.

Schoemaker, P. J. "The Expected Utility Model: Its Variants, Purposes, Evidence and
Limitations" Journal of Economic Literature, 20(1982):529-563.

Sandmo, A. , "On the Theory of the Competitive Firm Under Price Uncertainty",
American Economic Review, 61(1976):629-62.

Schluter, M. G. and T. D. Mount, "Some Management Objectives of the Peasant
Farmer: An Analysis of Risk Aversion in the Choice of Cropping Patterns-
Surat District, India" , Journal of Development Studies, 12(1976):246-61.

Schultz, T., Transforming Traditional Agriculture, New Haven, Yale University
Press, 1964.

Schurle, W. B. , and B. L. Erven, "Sensitivity of Efficient Frontiers Developed for
Farm Enterprise Choice Decisions" , American Journal of Agricultural
Economics, 61(1979):506-11.

Scott, J. T, and C. B. Baker, "A Practical Way to Select An Optimum Farm Plan
Under Risk", American Journal of Agricultural Economics, 54(1972):657-60.

Sengupta, J. K. , "Safety First Rules Under Chance-Constrained Linear Programming"
Operations Research 17(1969): 112-31.

Shwedel, K. , Marketing Problems of Small Farm Agriculture: A Case Study of the
Costa Rican Potato Market, Ph.D. Dissertation, Michigan State University,
East Lansing, 197 6.

Silva, Alvaro, Evaluation of Market Reform: Corabastos-Bogota, Ph. D. Thesis,
Michigan State University, East Lansing, 1976.

Simpson, W. , and M. Kapitany, "Off-farm Work Behavior of Farm Operators,
American Journal of Agricultural Economics, 65(1983):801-805.

Smale, M., Z. H. W., Kaunda, H. L. Makina, M. M. M. K., Mkandawire, Farmers’
Evaluation of Newly Released Maize Cultivars in Malawi: A Comparison of
Local Maize, Semi-F lint and Dent Hybrids, CIMMYT, Lilongwe, 1993.

Smale, M. , Chimanga Charrrakolo, Hybrids, and Composites: An Analysis of Farmers
Adoption of Maize Technology in Malawi, CIMMYT, Lilongwe, 1991.

Small, L. E. and I. C. (hrrruthers, Farmer Financed Irrigation: The Economics of
Reform, Cambridge University Press, Cambridge, 1991.

Sumner, D. , "The Off-farm Labor Supply of Farmers" , American Journal of
Agriculture Economics, 64(1982):499-509.

 

237

Talpaz, H. and J. W. Mjelde, "Crop Irrigation Scheduling via Simulation-Based
Experimentation" , Western Journal of Agricultural Economics, 13(1988): 184-
192.

Tauer, L. W. , "Target MOTAD, "American Journal of Agricultural Economics,
65(1983):606—10.

Taylor, C. R. , "Risk Aversion Versus Expected Profit Maximization with Progressive
Income Tax", American Journal of Agricultural Economics, 68(1986):136-145.

Telser, L. G., "Safety First and Hedging", Review of Economic Studies, 23(1955):1-
16.

Thompson, K. and P. B. R. Hazel] "Reliability of Using Mean Absolute Deviation to
Derive Efficient E-V Farm Plans", American Journal of Agricultural
Economics 54(1972):503-506.

Tirnmer, C. Peter, W. P. Falcon and Scott R. Pearson, Food Policy Analysis,
Baltimore: Johns Hopkins University Press, 1983.

Tobin, J. "Liquidity Preferences as Behavior Towards Risk" Review of Economic
Studies, 25(1958):65-86.

United States Department of Agriculture, Survey of Irrigation in Eight Asian Nations,
Foreign Agricultural and Economic Report No. 165, Washington, DC. , 1981.

Von Neumann, J. , and O. Morgenstern, Theory of Games and Economic Behavior,
Princeton, New Jersey, Princeton University Press, 1953.

Watts, M. J ., Larry J. Held and Glenn A. Helmets, "A Comparison of Target
MOTAD to MOTAD", Canadian Journal of Agricultural Economics,
32( 1984): 175-85.

Weimar, M. R. and A. Hallan, "Risk, Diversification, and Vegetables as An
Alternative Crop for Midwestern Agriculture", North Central Journal of
Agricultural Economics, lO(1988):75-91.

Whitrnore, G. A. , "Third Degree Stochastic Dominance" , American Economic
Review, 60(1970):457-459.

Wolgin, J. M., "Resource Allocation and Risk, A Case Study of Smallholder
Allocation in Kenya" , American Journal of Agricultural Economics,
57(1975):622-630.

World Bank, Poverty and Hunger: Issues and Options for Food Security in Developing
Countries, Washington, DC, 1986.

World Bank, Malawi Food Security Report, Report No. 815l-MAI, November 1989.

238

World Bank, Growth Through Poverty Reduction, Report No. 8140-MAI, Lilongwe,
March 1990.

Yaron, W. , and A. Dinar, Optimal Allocation of Farmer Irrigation Water During
Peak Seasons, American Journal of Agricultural Economics, 64(1982)681-689.

Yassour, J ., D. Zilberman, and G. C. Rausser, "Optimal Choices Among Alternative
Technologies with Stochastic Yield" , American Journal of Agricultural
Economics, 63(1981):718—723.

 

 

MICHIGAN STATE UNIV. LIBRARIES
llHI”WWIWWWIVWIHWIWIWWI“"WI
31293010468316