ufiwv
4“...

.3
ga
2.;

..
Ill. 1.5.1659. .25..
.Jiathiaiil.‘!l. 3“
r

 

 

LIBRARY
Michigan State
University

 

 

This is to certify that the
thesis entitled

AN ASSESMENT OF TOMATO PRICE VARIABILITY IN
LUSAKA AND ITS EFFECTS ON SMALLHOLDER FARMERS

presented by

Mukwiti Nchooli Mwiinga

has been accepted towards fulfillment
of the requirements for the

MS. degree in Agricultural Food and Resource
Economics

 

 

    

 

ajor‘ifii'fessor sS a[ature
14th Septemberaz

Date

MSU is an Afiinnative Action/Equal Opportunity Employer

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\ 5/08 KzlprojIAocatPrasIClRC/Dateouejndd

 

 

AN ASSESSMENT OF TOMATO PRICE VARIABILITY IN LUSAKA
AND ITS EFFECTS ON SMALL HOLDER FARMERS

By

Mukwiti Nchooli Mwiinga

A THESIS

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

MASTERS OF SCIENCE
Agricultural, Food and Resource Economics

2009

ABSTRACT

AN ASSESSMENT OF TOMATO PRICE VARIABILITY IN LUSAKA AND ITS
EFFECTS ON SMALLHOLDER FARMERS

Mukwiti Nchooli Mwiinga
This paper discusses the structure and operation of the tomato subsector in Lusaka
(Zambia), establishes the level of price variability for tomatoes in Lusaka’s Soweto
market, and assesses the impact of tomato price variability on returns to tomato
production. Price variability determination involved analysis of the coefficient of
variation, conditional variance and the ratio of the mean absolute positive to negative
price prediction errors. These results were compared with four other wholesale markets in
Costa Rica, Taiwan, Sri Lanka, and the United States of America (Chicago). These other
countries were chosen to capture a wide range of supply chain development, as proxied
by purchasing power parity Gross Domestic Product (PPP GDP). The study revealed that
(a) PPP GDP is strongly negatively (positively) associated with price variability
(predictability), and (b) Zambia has the lowest PPP GDP, highest price variability, and
least tomato price predictability. Monte Carlo simulation analysis was then conducted to
establish the effect that three different scenarios would have on the tomato farmers’ net
returns. Increased sales frequency reduces the variability of expected price but has no
recognizable impact on the variability of profits. Supply chain improvements also
reduced the variability of prices. The production of high quality tomatoes has very
significant effects on returns to farmers. Some policy implications drawn include, the
need to establish formal grades and standards, investment in cold chain systems and

general improvement in the traditional wholesale and retail market infrastructure.

DEDICATION
This thesis is dedicated to my family: my late father, Elijah Musernbwa Mwiinga, my
mother, Margaret Duntuula Mwiinga, my siblings Malita, Maelo, Mugwagwa and
Mutinta, and my nieces Chiedza and Duntuula. Special dedication goes to my dearest
father who always insisted on hard work and discipline in school, and to my mother for
all her love and support during my studies, but most importantly, for being the

‘superwoman’ in my life.

iii

ACKNOWLEDGEMENT
I would like to express my sincere thanks and appreciation to all those who helped me
finish my degree at Michigan State University (MSU). I am especially grateful to my
major professor, Dr. David Tschirley, for his support throughout my masters program.
His guidance, encouragement and understanding provided an ideal atmosphere for the
timely and proper completion of this work. My thanks also go to the other two members
of my thesis committee - Dr. Hamish Gow from the Department of Agricultural and
Resource Economics and Dr. Randy Beaudry fiom the Department of Horticulture, for
their valuable comments and suggestions. I am also thankful to Dr. Cynthia Donovan for
her assistance in selecting the courses I took during my masters program and her

preliminary ideas on my thesis.

Special thanks go to the USAID Initiative for Long-term Training and Capacity Building
Program (U ILTCB) administered through the Bean/COWpea Creative Research Support
Project (CRSP) at MSU for the financial support throughout my masters program. From
the Bean Cowpea CRSP, I am thankful to Dr. Irvin Widders, Dr. Mywish Maredia and
Mr. Ben Hassankhani for the important administrative role they played prior to and

during my masters program.
I am also thankful to the University of Zambia, School of Agricultural Sciences for

nominating me to participate on the UILTCB program and the Staff Development Office

for their fellowship support during the research phase of my program. Special thanks go

iv

to the Dean, School of Agricultural Science, Dr. Judith Lungu, and Dr. Olusegun

Yerokun for their leadership, facilitation and support throughout my studies.

My Sincere thanks also go to the Food Security Research Project (FSRP, Zambia) which
matched my research funds for the successful administering and completion of my
tomato survey, and also provided me with additional data for my research (tomato price
collection data and urban survey consumption data) and research support; field and office
resources and facilities. I am greatly indebted to Dr. Mike Weber who had his office
partitioned to create office space for me. I am also thankful to all the other FSRP staff
for the various ways in which they were of assistance to me, with special thanks going to
Mr. Munguzwe Hichaambwa, Mr. Kennedy Malarnbo and Esnart Musukwa, who I

closely worked with during my survey and data analysis.

I’d like to also thank all the other faculty in the Department Of Agricultural, and Food
Resource Economics at MSU for their contribution to my professional deve10pment
during my stay at MSU; my classmates and fiiends who made my stay at MSU happier
and memorable - I am specially thankful to Lillian and Sindi Kirimi and their two
daughters, Eric Bailey, Alda Tomo and Keneilwe Kgosikoma; and my fellow UILTCB
candidates who were like my family away from home, especially Mwape Malunga,

Dingiswayo Banda and Patrick Ofori.

Last but not the least, I am thankful to my God and Savior, Jesus Christ, for the blessings

he continues to bestow on my life.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... ix
LIST OF FIGURES ......................................................................................................... xi
KEY TO ABBREVIATIONS ........................................................................................ xii
CHAPTER 1 ....................................................................................................................... 1
INTRODUCTION .............................................................................................................. 1
1.1 Background ........................................................................................................ 1
1.1.1 The Situation in Zambia ............................................................................ 4
1.2 Objectives of the Study ..................................................................................... 7
1.3 Organization of Thesis ....................................................................................... 8
CHAPTER TWO ............................................................................................................. 10
TOMATO PRODUCTION AND MARKETING SYSTEM SERVING LUSAKA ..10
2.1 Data ................................................................................................................... 10
2.1.1 Urban Consumption Survey Data .......................................................... 10
2.1.2 Tomato Wholesale and Retail Price and Quantity Data ..................... 11
2.1.3 Data on Procurement Systems ................................................................ 12
2.2 Methods ............................................................................................................. 13
2.3 Fresh Produce in Consumer Budget Shares .................................................. 15
2.4 The Structure of the Tomato Production and Marketing System Serving
Lusaka ........................................................................................................................... 19
2.4.1 Overview ................................................................................................... 20
2.4.2 The “Traditional” Sector ........................................................................ 22
2.4.3 The ‘Modern’ Sector: Supermarkets and Processors .......................... 34
2.5 Price Behavior .................................................................................................. 41
2.5.1 Weekly Wholesale Prices in Soweto Market ......................................... 41
2.5.2 Weighted Average Prices by Marketing Channel ................................. 42
2.5.3 Tomato Wholesale and Retail Prices ...................................................... 43
2.5 Summary and Conclusions ............................................................................. 48
2.5.1 Importance of Tomatoes .......................................................................... 48
2.5.2 The Tomato Subsector ............................................................................. 48
CHAPTER 3 ..................................................................................................................... 52

TOMATO PRICE VARIABILITY AT WHOLESALE LEVEL: COMPARING
SOWETO MARKET (ZAMBIA) WITH OTHER WHOLESALE MARKETS

ACROSS THE WORLD ................................................................................................. 52
3.1 Factors Influencing Price Variability and Predictability ............................. 52
3.2 Hypothesis Testing ........................................................................................... 60
3.3 Data and Methods ............................................................................................ 61

3.3.1 Data ........................................................................................................... 61
3.3.2 Methods ..................................................................................................... 63
3.4 Results ............................................................................................................... 66

vi

3.4.1 Variability and Predictability of Prices ................................................. 66

3.4.2 ‘ The Problem of Predicting Sharp Price Declines .................................. 70
3.5 Summary and Discussion ................................................................................ 72
3.5.1 Tomato Seasonality of Supply ................................................................. 76
3.5.2 Tomato Supply Shocks ............................................................................ 79

3.5.3 Random Fluctuations in the Quantities of Tomatoes Arriving in the
Market 80

CHAPTER 4 ..................................................................................................................... 83
MONTE CARLO ANALYSIS OF CONDITIONAL AND UNCONDITIONAL NET
RETURNS TO TOMATO PRODUCTION .................................................................. 83
4.1 Household Survey ............................................................................................ 85
4.2 Price Data ......................................................................................................... 87
4.3 Overview of Monte Carlo Analysis ................................................................ 88
4.4 The Monte Carlo Model .................................................................................. 90
4.5 Results ............................................................................................................. 103
4.5.1 Distributions of Farmer Profits ............................................................ 103
4.5.2 Simulation Results for the Different Scenarios ................................... 104
4.6 Chapter Summary and Conclusion .............................................................. 114
CHAPTER 5 ................................................................................................................... 1 18
CONCLUSION .............................................................................................................. 1 18
5.1 Summary of key results ................................................................................. 119
5.1.1 Importance of Tomatoes ........................................................................ 119
5.1.2 The Tomato Subsector ........................................................................... 119
5.1.3 Tomato Price Variability and Predictability ....................................... 123
5.1.4 Baseline and Different Scenarios on Net Returns to Tomato
Production .............................................................................................................. 126
5.2 Contributions and Limitations of the Study ................................... . ............ 128
5.3 Future Research ............................................................................................. 131
5.4 Policy Implications and Recommendations ................................................. 132
APPENDICES ................................................................................................................ 1 36
APPENDIX 1. ................................................................................................................. 137
Checklist for Interview with FFV Procurement Managers for Supermarkets and
FFV Processors ............................................................................................................... 137
APPENDIX 2. ................................................................................................................. 139
Full Wholesale Tomato Price Prediction Regression Results .................................... 139
APPENDIX 3. ................................................................................................................. 144
Graphs of Price Prediction Residuals .......................................................................... 144
APPENDIX 4. ................................................................................................................. 149
Tomato Survey Instrument ........................................................................................... 149
APPENDIX 5. ................................................................................................................. 180
Distribution of Sampled Farmers ................................................................................. 180
APPENDIX 6. ................................................................................................................. 181

Baseline Distributions for the Random Variables Cost per Hectare and Yield ....... 181

APPENDIX 7. ................................................................................................................. 186
Histograms of Farmer Profits per Hectare under Different Scenarios .................... 186
BIBLIOGRAPHY .......................................................................................................... 206

viii

LIST OF TABLES

Table 2.1: Budget Shares for all Food Items Purchased by Households, in Four Cities of
Zambia ............................................................................................................................... 16

Table 2.2: Budget Share of Different FF V items in Overall FF V Purchased by
Households in Four Cities of Zambia ................................................................................ 17

Table 2.3: Budget Share of Different F FV Items in Overall FFV by Expenditure Quartile
for Households in Lusaka .................................................................................................. 19

Table 2.4: Key Characteristics of Tomato Production Areas Supplying Lusaka in 2007 .23
Table 2.5: Retail Outlet Market Shares on Overall Food (Lusaka) ................................... 32
Table 2.6: Retail Outlet Market Shares for all FFV Purchases by Income Quartile ......... 33

Table 2.7: Retail Outlet Market Shares for Tomato Purchases by Expenditure Quartile..34

Table 2.8: Weighted average tomato prices by market channel ........................................ 43
Table 2.9: Mean Tomato Prices for Wholesale and Retail Outlets in Lusaka (January
2007 to July 2008) ............................................................................................................. 46
Table 3.1: GDP Figures for Zambia and Other Selected Countries (Purchasing Power
Parity Terms) ..................................................................................................................... 61
Table 3.2: Description of Data Used in the Analysis of Tomato Price Variability ........... 62

Table 3.3: Yearly and Mean Coefficient of Variation of Nominal Tomato Prices in
Selected Countries ............................................................................................................. 67

Table 3.4: Yearly and Mean Conditional Variance of Nominal Tomato Prices in Selected
Countries ............................................................................................................................ 68

Table 3.5: Mean Absolute Values of Positive and Negative Tomato Price Forecast Errors

............................................................................................................................................ 7 1
Table 4.1: Results of t-test for Difference in Means .......................................................... 96
Table 4.2: Farmer Characteristics Based on Selected Variables ....................................... 98

Table 4.3: Basic Information on the Structure of Baseline Monte Carlo Simulation Model
.......................................................................................................................................... 1 00

Table 4.4: Distributions for Cost/ha ................................................................................. 101

ix

Table 4.5: Distributions for Yield .................................................................................... 102

Table 4.6: Confidence Intervals for the Profits per Hectare Variable in the Baseline

Model ............................................................................................................................... 103
Table 4.7: Baseline Results for Simulation Analysis ....................................................... 104
Table 4.8: Scenario on Increased Sales Frequency .......................................................... 106
Table 4.9: The Effect of Increased Sales Frequency on Tomato Profits ......................... 107
Table 4.10: Supply Chain Improvements ........................................................................ 109
Table 4.11: The Effect of Supply Chain Improvements on Tomato Profits .................... 110
Table 4.12: Production of Low and High Quality Tomatoes ........................................... 112
Table 4.13: The Effect of Tomato Quality on Tomato Profits ........................................ 1 12

LIST OF FIGURES
Figure 1.1: Geographical Location of the Republic of Zambia ........................................... 4
Figure 2.1: Channel Map for Tomato System Serving Lusaka ......................................... 21

Figure 2.2: Monthly Soweto Wholesale Tomato Prices January 2007 to July 2008 ......... 27

Figure 2.3: Weekly Soweto Wholesale tomato prices January 2007 to July 2008 ............ 41
Figure 2.4: Tomato Pricing at Wholesale and Retail Level ............................................... 44
Figure 2.5: Price Margin for Chilenje Retail Market ......................................................... 47
Figure 3.1: Mean Conditional Variance for Zambia and Four Selected Countries ........... 70

Figure 3.2: Comparison of the Ratio of the Absolute Mean Negative Errors to the
Positive Errors and the PPP GDP by Selected Countries .................................................. 72

Figure 3.3: Comparison of the Coefficient of Variation and PPP GDP by Selected
Countries ............................................................................................................................ 73

Figure 3.4: Comparison of Conditional Variance and PPP GDP by Selected Countries ..74

xi

KEY TO ABBREVIATIONS

FFV

FDI

FSRP

GDP

Ksh

MACO

M-W-F

NRDC

PPP

RNPE

SSA

UCS

UILTCB

UMDP

UN

USD

US/U SA

USAID

ZEGA

ka

Fresh Fruits and Vegetables

Foreign Direct Investment

Food Security Research Project

Gross Domestic Product

Kenyan Shilling

Ministry of Agriculture and Cooperatives
Monday-Wednesday—Friday

Natural Resources Development College
Purchasing Parity Prices

Ratio of the mean absolute values of negative to positive errors
Sub Saharan Africa

Urban Consumption Survey

USAID Initiative for Long Term Capacity Building
Urban Markets Development Program

United Nations

United States Dollar

United States of America

United States of America

Week

Zambia Export Growers Association

Zambian Kwacha

xii

CHAPTER 1
INTRODUCTION

1.1 Background
Research and programmatic activity on the horticultural sector in Africa over the past 15
years has been dominated by two issues: increasing horticultural exports and the
influence of emerging supermarkets on horticulture trends. Early in the period, Kenya’s
success in exporting fresh produce to Europe led to a large body of research documenting
the process and assessing its effects. For instance Jaffee (1995) investigated the
organization and development of a dynamic Afiican export oriented sector, specifically,
Kenya’s horticultural exports. Other documented research bring to light the recent
developments in Sub Saharan Africa horticulture exports and the success story in
Kenya’s horticulture sector (Swemberg 1995; Kimenye 1995; Stevens and Kennan 1999;

Dolan et a1. 1999; Harris et al. 2001; Minot and Ngigi 2002).

With the success that has been recorded in Kenya’s horticultural export sector, this has
also led to many programmatic initiatives across the continent to help countries exploit
what was seen as a rapidly growing and potentially very lucrative market. In Zambia for
instance, Foreign Direct Investment (F DI) has been instrumental in increasing exports of
horticulture and floriculture products in recent years. Much of the investment has gone
into the transfer of skills and knowledge, the introduction of new varieties of flowers and
vegetables, and made local farmers more familiar with the use of new pest control
methods and irrigation. For instance, the Natural Resources Development
College/Zambia Export Growers’ Association (NRDC/ZEGA) was set up mainly by

exporters, most of them foreign firms, in partnership with the government of Zambia.

Through this Trust, farmers are educated on the safe use of agricultural chemicals,

pesticides and herbicides, and on personal and consumer safety (United Nations, 2006).

The second main focus of research on African horticultural sectors has been on the rise of
supermarket chains. These chains have been seen as the leading edge of globalization in
developing country food systems, and concerns have been raised about the ability of local
retailers to compete, and also about the possible exclusion of smallholder farmers from
these new supply chains (Weatherspoon and Reardon 2003; Humphrey 2007; Reardon

and Berdegue 2002; Reardon and Timmer 2006).

Both strands of work — on horticultural exports and on the rise of supermarkets - have
made important contributions to our understanding of Afiican horticultural sectors.
Exports have been a major and continuing success story in Kenya, and other countries,
such as Cote d’Ivoire have also made some progress in developing these sectors.
Supermarkets have also expanded fairly strongly in some Afiican countries, and represent

a potentially important force of change.

Though these two strands of work have highlighted important aspects of current fresh
produce systems in Africa, they both miss two fundamental facts. First, the vast majority
of fresh produce in the continent is purchased by domestic consumers, not foreign buyers.
For example, Tschirley et a1 (2004) show that, in Kenya, during the period 1997 to 2000

retail domestic sales of vegetables accounted for 52% (valued at Kenya Shilling

(Ksh')7.5 billion) of total vegetable production, and vegetables that were retained on the
farm accounted for 36% (Ksh 5.2 billion) while only 12% (Ksh 1,7 billion) of domestic
production went to export sales. Yet Kenya is the foremost African success story in fresh
produce exports; in other countries of the continent, the dominance of the domestic over
the export system is even more accentuated. Second, within the domestic system, the
“traditional” systems carry the vast majority of all fresh produce in all Afi'ican countries
except South Afiica. (Tschirley 2007; Humphrey 2006; Traill 2006; Minten 2007). Even
though there could be a steady rise in the volumes of horticultural sales passing through
non-traditional channels such as supermarkets, many of these authors suggest that the

market shares of traditional channels are likely to remain high for many years in Africa.

Despite the current widespread use of traditional horticultural retail channels, they have
received very little public- or private investment since independence, and this lack of
investment is a major problem, causing congestion, unsanitary conditions and high costs
(Hichaambwa and Tschirley, 2006). High price volatility is a major challenge in all fresh
produce systems due to their perishable nature. Even more challenging in traditional
system is the lack of cold chains, little or no timely market information and the general
absence of coordination mechanisms to regulate the flow of product to the market (World

Bank 2007).

 

' The mean exchange rate to the US $ for the four year period between 1997 and 2000 was KSH 66 (KSH
59, KSH 60.5, KSH 70 ND KSH 76 for 1997, 1998, 1999 and 2000 respectively. www.0anda.com

Given these problems faced by traditional systems, if not vigorously addressed, they will
only become worse over time, due to rapid urbanization and income growth that fuels

even more rapid growth in demand in mban areas.

1.1.1 The Situation in Zambia
The republic of Zambia is a landlocked country located in Southern Africa bordered by
eight countries namely: Mozambique, Malawi, Tanzania, Democratic Republic of Congo,
Angola, Namibia and Zimbabwe (Figure 1.1). The country has a population estimated at
12.5 million with 65% being rural population and 35% urban population, and has a Gross

Domestic Product (GDP) per capital of U331 ,2232.

Figure 1.1: Geographical Location of the Republic of Zambia

 

Source: http://www worldatlac m.../.. L‘ ’........trys/africa/zm.htm
In Zambia, nearly 90% of all fresh produce marketed in Lusaka3 flows through traditional
retail channels, specifically the open air markets and street vendors and other informal

traders operating outside the market, while modern retail channels such as supermarket

 

2International Monetary Fund (IMF) publications.

http://www.irnf.orgzextemal/pubs/fi/weo/2008/02/weodata
3 Lusaka is the capital city of Zambia and has the largest FFV wholesale and retailing system.

chains and independent supermarkets hold combined shares of less than 10% (Food
Security Research Project Urban Consumption Survey, 2007). This clearly tells us that
the traditional sector dominates the fresh produce system as in most of Sub Saharan

Afiica (SSA).

In many SSA countries, there has been rapidly rising share of urban population in total
population. According to the Population Division of the Department of Economic and
Social Affairs of the United Nations Secretariat, over the past few decades and in the next
to come, the percent of urban population has been and will continue to rise steadily
compared to the rural population which is actually decreasing“. However, in Zambia this
has not exactly been the trend. Over the period between 1980 to 2030, the percent of
urban population had initially been increasing, then it begun to decrease in the 1990’s and
then steadily rising after 20055. The decreasing annual urban population trend is
attributed to the investments made in the mining sector which saw a good number of

people moving to the rural mine areas for employment.

Considering the overall increase in the urban population in SSA, the traditional marketing
channels in African horticultural sectors are now subject to heavy pressures for change.
In Zambia, urban populations are growing rapidly and therefore the traditional
horticulture systems need substantial investment. Since urban marketing infrastructure in
most of the continent, Zambia included, has received very little investment in recent

decades, the result has been often chaotic, unsanitary, and high-cost marketing systems

 

4 World Urbanization Prospects: The 2007 Revision Population Database

http://esa.unorgZunup/QZkOdata.asp
5 ht_tp://esa.unorgZunup/QZkOdataasp

that don’t serve the interests of farmers or consumers very well (Hichaambwa and

Tschirley, 2006).

Soweto market in Lusaka is the largest wholesale and retail center for fresh fruits and
vegetables (FFV) in the country. Located in the center of the city, it is a commercial hub
for F FV and a wide assortment of other food items such as dry cereals, pulses, and tubers,
among others. Despite the huge amounts of F FV and other food items it handles, this
market has for a long time been in a poor state. It has poor and limited sanitation, a poor
waste management system and a poor drainage system. Even though the local council
authority collects market stall levies from the operators in this market, there has been
little investments made to improve it. Coupled to its physical inadequacy is the absence
of market information, and the lack of formal grades and standards. (Hichaambwa and

Tschirley, 2006, Typsa Consulting Engineers and Architects, 2004)

In an attempt to address some of the concerns in Soweto market, the European Union, in
collaboration with the Ministry of Local Government and Housing is currently investing
over 16 million Euros into a program called the Urban Markets Development Program
(UMDP). Among other things, the UMDP has focused on the construction of improved
physical infrastructure in selected markets of Lusaka, Ndola and Kitwe cities. In Lusaka,
Soweto market is one of the markets that has benefited from this program (Hichaambwa
and Tschirley, 2006). This program is currently ongoing in all selected markets and
works in Lusaka’s Soweto market still continue. Despite the investment made by UMDP

in Soweto market, it is not clear how meaningful a contribution the program will make

towards lowering the costs, encouraging higher quality and better price predictability,
among other things, for tomatoes and other FFV, and generally improving wholesaling

and retailing of fresh produce in the market.

This paper shall focus on the wholesaling and retailing of tomatoes in Lusaka’s Soweto
market. Among all FFV, tomatoes have the second largest share in both production and
consumption in Zambia, following rape, (F SRP UCS data, 2007). Tomatoes are therefore
one the most widely consumed flesh fruits and vegetables. However, farmers, traders,
and consumers of these tomatoes are faced with tremendous price variability flom day to
day and also within days. Given the high level of variability of tomato prices, two
important questions arise. Firstly, what is the effect of this variability on the riskiness of
rettu‘ns to farmers? And secondly, can market information lead to improved decision
making that raises and stabilizes returns? This paper shall address these two questions.
1.2 Objectives of the Study

The overall objective of this study is to evaluate the price variability of tomato in
Lusaka’s Soweto market and to assess the effects of different production and marketing
strategies on farmers’ performance. Soweto market accounts for the lion’s share of

wholesale flesh produce transactions in Lusaka.

The specific objectives of the study are:

1. To identify the level of price variability for tomato in Soweto market and evaluate

how this compares with other markets around the world.

2. To determine the impact of price variability on the current level and variability of
farmer returns to tomato production.
3. To assess the effects of alternative production and marketing strategies, and “generic”

supply chain improvements on the variability of price and returns to farmers.

1.3 Organization of Thesis
The thesis organization is as follows: Chapter 2 gives a detailed overview of the tomato
production and marketing system serving Lusaka. The data and methods of analysis used
are also discussed. A tomato subsector channel map is presented with a discussion on the

various actors in the subsector.

Chapter 3 presents the hypothesis, an analysis, results and discussion of the tomato price
variability at wholesale level where a comparison of Soweto market with some other
wholesale markets in other parts of the world was made. Tomato price variability in
wholesale markets of the US, Taiwan, Costa Rica and Sri Lanka was analyzed and

compared with that of Zambia’s Soweto market.

Chapter 4 presents the data used, results and discussion on the various Monte Carlo
analyses conducted. Both the conditional and unconditional distribution of the tomato
growers net returns flom tomato production are looked at. The conditional profits
discussed are based on the tomato grower’s production and marketing decisions and on
the use of certain market information, while the unconditional profits are based on the

tomato prices they observe in the market.

Chapter 5 concludes the thesis with a presentation of a summary of the study, policy
implications, contributions and limitations of the study, and suggestions for future

research.

CHAPTER TWO
TOMATO PRODUCTION AND MARKETING SYSTEM SERVING LUSAKA

This chapter provides a broad and detailed examination of the tomato production and
marketing system serving Lusaka. It starts by explaining the broad array of data and the
methods used. It then uses these data to examine flesh produce in consumer budget
shares across four cities of Zambia. Next, it focuses on Lusaka, presenting an overview
of the structure of the tomato production and marketing system for the city, organized
around a detailed tomato channel map that brings together data and information flom
many sources. Key points about the organization of the sector are highlighted in this
section, and are examined in more detail in subsequent sections. After the overview is a
discussion of the entire vertical supply chain for the “traditional” sector followed by the
“modern” sector. In closing, price behavior at the farm, wholesale, and retail levels is
examined.
2.1 Data
The data used in this chapter comes flom three sources: the Food Security Research
Project6 Urban Consumption Survey (FSRP-UCS), the FSRP wholesale and retail price
and quantity data, and data on the F FV procurement systems adopted by some selected
retail outlets and FFV processing firms in Lusaka.
2.1.1 Urban Consumption Survey Data

The UCS was conducted in 2007 and it contained household consumption data flom
urban consumers in Kitwe, Mansa, Lusaka and Kasarna cities, all in Zambia. This survey

was collected from a sample size of 2,160 urban consumers who were sampled using a

 

6 The Food Security Research Project (F SRP) has operated in Zambia since 1999, with funding flom US.
Agency for International Development/ Zambia and, recently, flom the Swedish International Development
Agency. Over the past decade it has collected various household- and market level data sets in
collaboration with local organizations; some of those data sets are used in this thesis.

10

randomized cluster sample design. This data contains specific information on the various
FFV, other food and non-food items purchased and consumed by the household; the
value of consumption for all the foods purchased, and the primary retail outlet in which
each item was purchased. In total, there was data collected on 37 FFV and food items
such as rape, tomato, onion, cabbage, cassava leaves, sweet potato leaves, pumpkin
leaves, bananas, mangoes, oranges, apples and beans, and nine non-food items such as
fire wood, paraffin, batteries and vaseline jelly. In addition to this data on the FF V, food
and non food items, the survey also collected data on household expenditures, the
households’ participation in urban agriculture (horticultural crop production and livestock

production) and the households’ food security levels.

2.1.2 Tomato Wholesale and Retail Price and Quantity Data
FSRP market reporters collect price data at wholesale and retail, and quantity data at
wholesale, for tomato, rape, and onions every Monday, Wednesday and Friday. This
quantity data captures total volumes of tomatoes (and the other two crops) moving
through Soweto market, while the price data is collected at Soweto and selected retail
outlets. Soweto market is supplied with tomatoes flom over 150 areas from Lusaka and
Central provinces. Quantity data includes information on the area of origin and the size
(number of crates) of every lot entering the market. By “area of origin” we refer to a
production area at the sub-district level as identified by farmers and traders selling in the
market. A “lot” is defined as the set of crates belonging to an individual farmer or trader
whose tomatoes are being sold in the market. The Soweto wholesale price and quantity

data7 tracks enterings, starting and ending volumes of tomatoes in the market flom all the

 

7 Food Security Research Project: Tomato wholesale and retail price and quantity data.

11

supply areas. Three price observations are collected and recorded each hour, and the
mean of the three prices is taken as the hourly tomato price. This is particularly the case

for the price data on entering volumes.

Retail outlets where price data is collected are Shoprite (Cairo/Kafue roads), Spar (Down
town), and Melissa (Matero) supermarket chains, and Chilenje open air retail market. All
these data have been collected since January 2007; however, in generating the subsector
map, only data for the one year period January to December 2007 was used.
2.1.3 Data on Procurement Systems

Data on the procurement systems of selected retail outlets and F FV processors was
obtained flom interviews with the procurement managers of these institutions. The
interview guide used for these interviews is presented in Appendix 1. This guide took the
form of a checklist questionnaire with a combination of closed and open ended questions.
Open ended questions were included in order to get the broadest possible insights into the
nature of the procurement systems these actors have adopted. Some of the information
solicited flom the interview included what F FV they trade in, who their FFV suppliers
are, and more specifically, who their tomato suppliers are, the geographic origin of the
tomatoes flom their suppliers, the quantity and specification requirements of the
tomatoes, and their tomato pricing policy. The interview was conducted on three large
independent supermarket chains; Shoprite (Fresh mark), Spar and Melissa, and the two

main FFV processors in the country; F reshpikt and Rivonia.

 

8 Entering volumes are those entering the respective markets between 6am and 1 pm; starting volumes refer
to the volumes of tomatoes entering between the end of the previous day and 6am, while ending volumes
are those sitting in the market still unsold at noon each day.

12

2.2 Methods
With these three sets of data, two types of analysis were conducted. The first involved
ascertaining the importance of tomato among all the FFV and the second one involved
the analysis of the tomato subsector which also included some analysis of the

significance of the traditional retail sector.

The UCS data was used to ascertain the importance of tomato among all FFV, and also to
show the significance of the traditional retail sector. Based on household expenditure, the
first analysis involved calculating:

- budget shares of all food items purchased,

- budget share of F FV in overall FF V purchased, and

- budget shares of all FFV items for Lusaka by income quartiles
In determining the significance of the traditional retail sector, the analysis involved
calculating:

- retail outlet market shares for all food items purchased,

- retail outlet market shares for tomatoes by expenditure, and

- retail outlet market shares for all FF V by expenditure quartile groups.
In conducting the tomato subsector analysis, the tomato wholesale and retail price
collection data, the UCS data and the data on the FFV procurement systems were used.
These data provided information on:

- main actors in the tomato sub sector,

- volumes of tomatoes flom the various identified farm areas,

- volumes of tomatoes flom the retail outlets,

13

- various channels through which tomatoes pass through before they finally reach
the retail outlets,
- volumes of tomatoes which are handled by the traders and their sources,
- lot sizes of tomatoes flom the farmers, and
- type of first sellers of tomatoes in Soweto market; farmers or traders.
Volume data on supply areas was used to calculate total supply of tomatoes flom each
area and also the total supplies channeled through the various identified marketing

channels.

Data on. the lot sizes of tomatoes flom different supply areas was used to estimate the
relative size of the farmers flom these supply areas. The lot sizes flom the farmers were
then categorized into terciles. Based on where lot sizes flom a particular area fell in the
tercile groups, each supply area was categorized into three groups flom the largest to
smallest implied farm size. The strength of using this approach of categorizing the supply
areas is that it gives a good estimate of the size of the majority of famers in a given area.
However, the down side to this approach is that it may underestimate or overestimate
actual sizes of the farmers in the supply areas. For instance, several small farmers may
have been categorized as large farmers merely on the basis of a few large lot sizes of
tomatoes they delivered to the market, or conversely, a few large farmers may have been
consistently delivering small lot sizes of tomatoes very often and were subsequently
categorized as small farmers. In general, however, the implied farmer size and resulting

classification of production areas that emerged flom this exercise agree with the

14

perceptions of farmers and traders in the market regarding the farm structure in most

areas.

To understand the FF V procurement systems of the large independent supermarket chains
and FFV processors, interviews were conducted with the procurement managers of these

institutions.

For the subsector analysis, the UCS provided data on the retail outlets the consumers
purchase their tomatoes flom and the volumes of tomatoes purchased in each retail outlet,
and information on the tomatoes that were grown and consmned by individual
households and the tomatoes which were given to the households as gifts. This
information was obtained by summing up the quantities of tomatoes by each retail outlet
or source. The main output of the tomato subsector analysis was a subsector map which
shows all the main actors in the system and the total volumes of tomatoes in each

identified channel.

2.3 Fresh Produce in Consumer Budget Shares
Fresh fluits and vegetables are one of the most widely consumed food items among
households in Lusaka and the other three surveyed cities (table 2.1). In all four cities,
vegetables and fluits account for 12% of all purchases. In Lusaka, F FVs are fourth (taken
together) in budget shares after cereals/staples, meat/eggs and other foods.
In all four cities, tomatoes and onions are in first place, with an average share of 10% of
all expenditure on FFV (table 2.2). Clearly, tomatoes are a major FFV consumption item

and therefore have an important impact on households’ purchasing power.

15

Table 2.1: Budget Shares for all Food Items Purchased by Households, in Four

 

 

Cities of Zambia
All 4
Food Group Cities Kitwe Mansa Lusaka Kasama
----- Share in total food expenditure ------
Cereals/ staples 0.22 0.25 0.26 0.19 0.23
Meat, eggs 0.20 0.17 0.16 0.21 0.18
Other foods 0.17 0.14 0.16 0.20 0.15
Non-food items 0.13 0.13 0.13 0.13 0.14
Fish 0.10 0.09 0.11 0.10 0.12
Vegetables 0.07 0.10 0.09 0.05 0.08
Fruit 0.05 0.05 0.05 0.05 0.04
Legumes 0.04 0.04 0.04 0.04 0.04
Dairy 0.03 0.04 0.02 0.02 0.03

 

Source: Food Security Research Project Urban Consumption Survey Data 2007

16

Table 2.2: Budget Share of Different FFV items in Overall FFV Purchased by
Households in Four Cities of Zambia

 

 

Consumption Item All cities Kitwe Marisa Lusaka Kasama
Tomato 0.10 0.10 0.11 0.10 0.10
Onion 0.10 0.10 0.10 0.09 0.09
Rape 0.09 0.09 0.09 0.10 0.09
lmpwa 0.07 0.07 0.06 0.06 0.08
Cabbage 0.07 0.06 0.06 0.07 0.07
Sweet potato leaves 0.07 0.07 0.07 0.06 0.07
Pumpkin leaves 0.06 0.06 0.07 0.06 0.06
Bananas 0.06 0.06 0.06 0.07 0.05
Okra (lady‘s finger) 0.05 0.06 0.04 0.07 0.04
Oranges/ tangerines 0.05 0.05 0.05 0.05 0.04
Cassava leaves 0.05 0.05 0.07 0.02 0.04
Mangoes 0.04 0.04 0.05 0.03 0.04
Bean leaves 0.03 0.03 0.04 0.02 0.05
Lemons 0.03 0.04 0.03 0.04 0.02
Amaranthus (bondwe) 0.03 0.02 0.03 0.03 0.05
Avocado pear 0.03 0.03 0.02 0.02 0.04
Apples 0.03 0.03 0.02 0.04 0.01
Guavas 0.02 0.02 0.01 0.02 0.02
Green beans 0.01 0.01 0.00 0.03 0.01
Watermelons 0.01 0.01 0.00 0.01 0.00
Eggplant 0.01 0.01 0.00 0.01 0.00

 

Source: Food Security Research Project Urban Consumption Survey Data 2007.

Analysis of the budget shares of the different FFV items consumed by the households
over their total expenditure on FFV, by expenditure quartile was also conducted (Table
2.3). Using the data on all the food and non-food expenditure items, the households were
grouped into expenditure quartiles. These quartiles were calculated by first summing all
household expenditures for food and non-food items. Households were then ordered flom
the highest to lowest total expenditure then broken into four groups of equal size.
Quartile l is the least expenditure group and has a mean total expenditure of ZMK9 489,
700, while quartile 4 is the highest expenditure group with a mean of ZMK 3, 867, 700.

Quartile 2 is the second lowest expenditure group with an average income of ZMK 894,

 

9 The mean exchange rate to the USD during 2007 was (Zambian Kwacha) ZMK 4, 1 14; Source:
www.0anda.com

17

800 and quartile 3 is the second highest expenditure group with a mean expenditure of

ZMK l, 508, 205.

The results show that, tomatoes rank first in the first expenditure quartiles, while tied
with rape in first rank in the quartiles 2 through 4. Among the households in the third and
fourth expenditure quartiles, tomatoes had a budget share of 9% of the total F FV
expenditures of the household while the households in the income quartile 2 and 1 had
10% and 13% budget share of tomatoes respectively, over all FFV items. Both rape and
tomatoes have the largest budget share among the relatively poor households (rape forms
a very prominent part of relish eaten with nshima10 for these households), they both have
the same pattern across all the quartiles falling flom 13% to 9% for tomatoes, and 12% to
8% for rape. This basically shows the importance of tomatoes regardless of the

household income levels.

 

'0 Nsima is a maize meal pulp made flom maize flour and is the main staple consumed by households in
Zambia.

18

Table 2.3: Budget Share of Different FFV Items in Overall FFV by Expenditure
Quartile for Households in Lusaka

 

 

Expenditure Expenditure Expenditure Expenditure

Consumption Item quartile] quartile 2 quartile 3 quartile 4

Rape 0.12 0.10 0.09 0.08
Tomato 0.13 0.10 , 0.09 0.09
Onion 0.11 0.10 0.09 0.09
Cabbage 0.08 0.07 0.07 0.07
Chinese cabbage 0.01 0.02 0.01 0.01
Cassava leaves 0.02 0.02 0.02 0.02
Sweet potato leaves 0.06 0.06 0.06 0.05
Pumpkin leaves 0.06 0.06 0.06 0.05
Amaranthus (bondwe) 0.03 0.03 0.03 0.03
Bean leaves 0.01 0.02 0.02 0.02
Okra (lady's finger) 0.08 0.07 0.07 0.06
Impwa 0.06 0.07 0.06 0.06
Eggplant 0.00 0.01 0.01 0.02
Green beans 0.02 0.02 0.03 0.04
Bananas 0.05 0.06 0.07 0.07
Mangoes 0.03 0.03 0.03 0.03
Oranges/ tangerines 0.04 0.05 0.06 0.05
Apples 0.02 0.03 0.04 0.05
Avocado pear 0.01 0.02 0.03 0.03
Watermelons 0.01 0.01 0.01 0.02
Guavas 0.01 0.02 0.03 0.02
Lemons 0.04 0.03 0.04 0.04

 

Source: Food Security Research Project Urban Consumption Survey Data 2007
2.4 The Structure of the Tomato Production and Marketing System Serving
Lusaka
This section examines the structure of the tomato production and marketing system
serving Lusaka. This system is composed of tomato farmers categorized in three areas
based on the farmer types that dominate the area, tomato assemblers/processors, tomato

wholesalers, and a wide range of retailers.

Over 90% of tomato wholesale volume flows through the traditional sector, with less than
10% volumes flowing through the modern sector comprising Freshmark, which is a

formal wholesaler and processors Freshpikt and Rivonia.

l9

The retail sector is composed of both informal and formal actors. The informal system is
composed of open air markets and the “ka sector””, which refers to all small FF V
vendors, while the formal system is composed of the large independent supermarkets,
large chain supermarkets, mini marts and small super markets
2.4.1 Overview

Figure 2.1 presents a simplified channel map for the tomato system serving Lusaka.
About two-thirds of all tomato in Lusaka comes flom areas dominated by large and
medium size farmers. Also about three quarters of all volume is directly marketed by
farmers with less than one-fifth of these tomatoes first going through rural traders.
Travel times flom the production areas to Soweto are mostly under l 1/2 hours, with the
longest times being 4 hours. The market channel for tomatoes arriving into Lusaka is
therefore actually quite short. Freshpikt is the predominant FFV processor in Zambia and

it accounts for 8% of the tomatoes in the system all of which it produces on its own.

Over 80% of tomatoes flom farmers end up in Soweto market with less than 10% going
to Bauleni market. Soweto market clearly dominates as the main wholesale entity in
Lusaka. The processing and modern wholesaling sectors, dominated by Freshpikt
(Freshmark and Rivonia have extremely small shares) take less than 10% of the market.

In the retailing section, the traditional sector dominates with over 90% of the market.

 

u The “ka sector” refers to the informal retail outlets for FFV and these include market stands, market stall
vendors, mobile vendors, street vendors, ka table (small table stall), kantemba (small rudimentary shop)
and ka shop (kiosk) (FSRP Urban Survey Training Manual, 2007)

20

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m
. m m m m
m. d.
m m m w W m m W m Asarsocw 5. $1.3 acids. ._< :30 :33.
r
m
( o_ono_o:>>
w Ashe 05:50.0.
/ r
1o. % $9 $3 32.
W ( n .2525 N .occwco r .occwzo
m M
H \
m N 7 V // {on
m 3...: 1.5.26
F m up 22.5.; 1 / «nu acimououm
.- a .2228 ‘>_nEonu<
L -\nn—.
fl o .2226
\
K /\
_ A33. £63 area. 5.832..
i _ :93 Eau canar— ou2< Emu 83:90.2 na2< ..Enu =uEm

 

 

3.3:; Mariam 889$. Beach. .8.— maz 1:325 “fin ensur—

21

2.4.2 The “Traditional” Sector
The traditional wholesale sector is made up of Soweto and Bauleni markets which
together have an overall market share of 91% at this level. At retail, the traditional sector
has a 92% market share and is composed of the open air markets and the ka sector. These
results clearly Show how both the wholesale and retail traditional sectors dominate the

tomato subsector.

Soweto market is the main wholesale channel through which tomatoes pass before they
reach the various retail outlets. This market is supplied by a wide range of geographic
areas that include small, medium and large farm areas. Bauleni market on the other hand
is a small wholesale market that has much of its tomato supplied by farmers in small farm
areas, specifically flom Manyika in Chongwe district. Bauleni market is on the flom this
area to Soweto market, and as such, quite often farmers flom Manyika would opt to sell
their tomatoes in this market when they have smaller quantities which can easily be

purchased in this market, thus making proceeding to Soweto market unnecessary.

i. Production Areas
The FSRP price and quantity data base described earlier identifies 150 distinct areas that
supplies Lusaka with tomato during 2007. Of these, the twelve main geographical areas
that produce and supply tomatoes to Soweto market are Chalirnbana, Chisarnba, Choona,
Lusaka West, Makeni, Masansa, Manyika, Mkushi Farm Block, Mwaalumina,
Mwembeshi, Nkolonga and a special grouping of farmers flom Kapiri Mposhi district
(table 2.4). These twelve areas account for 68% of the tomato supplies that reached

Soweto market during 2007.

22

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.88 a; 88
>m2>tq< mo Samoa 02.5
32 5 33.5588 33m
8.5.850 :5 .88 ah? .momoov 832 05
3.80 98 £032 3 voted 5 0.8 808.3“ came
23 BE can me e8 cameo sea ”scene =oem 2.". one So 380 e826
boom 598800
ow nonfio>oz - 3:88
3E 52 war—3c anon mac—mm .mozooc 93 QB
8505 £5 .boom 33 8 05 5 Pa who—Ea
.32 - 3:58 085 32 2: E ofimo brown.
m8; woeebgoaoo .388 0.8 83m 06 “neon—Em 0mg 86 3." «mod 3.300 36820
modem
so: 88 Gaines:
eee coca SE 88 .838
538902 83 Kano:— ootm cone as 05
32 eee ewe assess... .88 a on aoaua as as
«on; 532932 women “magma $.08 ”mung“ owe—J ewe mo; £86 1356 3532
.momoow 5...
Ba an 2: E mecca
as» :5 nomooe
Sou beacon mo 5:08 02.5 as $88 Bo 38%
:3: war—3c gang team .noou >305 “mega
nonaofiom 8 .32 mo voted 98 flog Havana
v8; 8E 32 worse 8:3 302 me 39¢ :4 whm N: 2:6 8.354 $03 $34
gigginén use Awafinflmiioflalicalfimmllieai
@9533“ 22:9 mum—2533 58.3.”— .3 he Max:5— SA 5:32 uoxaa2
3....— ouaao>< 0:89 :32
Scene?
Sam 52.3,.—

 

 

Sen 5 873.5 warm—n95 26.2 552:5...— Sanh he moratoaoaaanu hov— ”~.N 03:.

23

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.woom
Eden—E 8 Sea “0980009
88 83 20.82 3 38:5 3:000
I 06:08 02.8 8.8 ward—u 23 no“ 06 8 0.8 808.5 202m 88m
$2 000880080 028 3.02 05 .«o 0008 ”808.5 093 82. SN 95.0 18800 20:22
80:02... 5w 08w 2:. 08 8
room 00980000 I 258 :23 :5 81.020 :0 9.903
2808 02.5 2&5 SOON w=< So 030.80 >805 8082a 08
m3 2 08: - 8888 02.8 33 808.8.“ 05 ”808.3.“ m0 3&2? 86 2.2 Sod .2085 8382—80
88 .3888
9 .33 mo 02808 02.5 32 80:00“. 0032 0:... 8 08 8085
am 8 0028800800 .530: 808m 08.8 808 803.2 =08m :1. $6 256 8200:.— 80202
8333030
0.88 m8>8 02800 2.5
Room «098080m 08 88 8m 05 2:3 8:000 =3
8 .32 I 82808 02.8 32 8 So 303% >803 “3820 08
mg 8 0008800800 .3800: 0.20m 808.8 08 8082 mo 89$ 5.. Sum 22 mod 328:2 20098032
.88 a? o. 88 832
mo cocoa 02.5 32 08 8
028.. no 8883 38838,. a
85552: :5 .88 3258
8 SON $98000Q 020 .82”
Edison mo 5:08 0288 3:000 32 08 8
moo; amE war—80 020m 8028: 0.8 80:88 808 “8088 288w omé pd Sod 880:2 8:832
Nam—JWGANV
003003—
025 0u20>< $5.88 0.8.:
3330.5 ban—6v 5:883; 02m .388..— 2282 00520.5 .3:

 

 

P250 N.N 035,—.

24

6:96 988 Began .833 05 a S 98 0:86 :38 033mg “mo—RSV. 05 mm fl

_

woofiboom Saw ban—gm 82:8. I .8._.Em suaflm bravom woo..— “venom

 

 

 

 

 

 

 

 

 

 

 

 

 

v.33 =a
o8; - - - - wed - é owns;
.omoow
5w 05 E 208 H23
.83 Eng m can n 5233
2 boom Bassoon mo 3308 3:26 :2on
8.59 :3: M39 9803 ”boom 05 E 1882
Honfioaom 8 09: mo 3:85 08 flog.“ “nee
0% 8E 32 mafia“. viva £22 “325$ 8332 EH 0: «86 8395 «585332
.88 Maegan
9 E2: Ba .88 3.53
E32 .88 598302 .mozoov Bow no“ 05 3834
9 33800 I 3:88 E can flog.“ 05 mo Eng
Nun; aura :wE wfihfi mafia $.32 “88 ”whoa—ham ems has mmd Rod 1350 E 30:55
.33 .9263.
Eng 8 boom “38302 028 no“ 05
I 3:88 035 "RE waist 50.8 €085 ufimo
me: @8580 8:3 8“ 3802 308 “308.5 ems VON. end Sod 3.550 «32on
fim§v
3339i
31.— omauo>< Amy—E2: 9:2.»
13:39.? 2.5sz baanoaaom «Em hon—nah .8132 85>??— 8.2

 

 

Pence QN 039—.

25

Monthly wholesale prices per kilogram of tomato in Soweto market for the period
between January 2007 and June 2008 were analyzed (figure 2.1). From the figure, over
the 19 month period it was observed that there are a number of high and low price
months and also sudden price drops which are a concern. The notable high price months
during this period are February to March 2007, October and November 2007, and
January 2008 and February 2008, while the low price months were around April to
August 2007, December 2007 and March 2008. A closer examination of the seasonality
during January through June in both years reveals that at the beginning of both years the
prices are fairly high and then there is a sudden price collapse. In 2007, the price collapse

occurred in April while the 2008 price collapse occurred in March.

26

Figure 2.2: Monthly Soweto Wholesale Tomato Prices January 2007 to July 2008

2000.00-

3 1500.00-
‘2‘
E
N
v
a
X
h

3. 1000.00-
0
.2
h
0.

500.00-

 

 

l T l l T T r If I l I I fi I l l l l 1
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL
2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 2008 2008 2008 2008 2008 2008 2008

Data

Source: Food Security Research Project — Tomato price data 2007/2008

Among the top twelve supply areas, those that supplied tomatoes to the market in both
high and low price months are Lusaka West, Manyika and Mwalumina. Farmers in areas
like Mkushi farm block, Nkolonga and Kapiri Mposhi district supplied their tomatoes
mainly in the high price months. In the low price months, the dominant supply areas were

Masansa, Chisamba, Choona, Mwembeshi, Makeni and Chalimbana.

Judging by the quantities of tomato that were supplied from the various supply areas in

April 2007, Masansa, Choona, Manyika and Lusaka West had the highest volumes of

27

tomatoes in that period (298mt, l82mt, 75mt and 7 Smt respectively) and accounted for
59% of the tomatoes on the market. On account of this, their supplies are likely to have

been the main cause of the April price collapse

In the case of the March 2008 price collapse, Choona and Masansa collectively supplied
the market with 49% of the tomatoes. Choona alone accounted for 20% and supplied the
market with 247 mt while Masansa supplied 174 mt. The supplies fiom the two areas are

to some extent largely responsible for the March price collapse.

As noted earlier, supply areas such as Mkushi farm block, Nkolonga and farmers in
Kapiri Mposhi district supplied the market with tomatoes mainly during high price
months. These supply areas are dominated by large farmers who generally have more
financial resources and farming knowledge than small farmers. The high price months
these farmers supplied their tomatoes in is indicative of a tomato crop grown in the rainy
season, during which production costs can be very high. These high production costs are
associated with high weed management requirements and more frequent pest and disease
outbreaks which require chemical applications for their management. Being large
farmers, it is easier for them to grow and manage a rain fed crop since they have more
financial resources to engage labor for weeding and buy chemicals for pest and disease
control. In addition to this, with the edge they have in farming knowledge, this puts them

in a better position to manage their crops well.

28

On the basis of the different supply areas and the different farmer types found in each
supply area, the channels through which tomatoes enter the system is presented in figure
2.1. Channels 1 through 3 represent tomatoes taken directly to the markets by farmers
from all 150 supply areas while channels 4 through 6 represent tomatoes that were first

sold to traders.

Channel 1 represents the flow of tomatoes fiom small farm areas into Soweto market.
Among the top twelve supply areas, channel 1 was made up of famers from Choona,
Manyika and Makeni. This channel has a 19% share of tomatoes entering Soweto market.
The Soweto data on the origin of tomato supplies shows that the majority of the farmers
in this channel are located in the lower deciles with only a few in the top two deciles.
Farmers in this area mainly supplied their tomatoes to Soweto market in the low price

' months of March to May 2007 and March of 2008.

The medium farm area is represented by channel 2 and among the top 12 supply areas
had farmers from Lusaka West, Mwembeshi, Chalirnbana and Mwalumina. A large
number of farmer observations are well distributed in all the deciles with the majority of
them lying in the 5th and 9th deciles. The farmers in this channel account for 28% of the
tomato volumes in Soweto market. This area supplied most of its tomatoes in the low

price months of May to September 2007.

The large farm area is represented by channel 3 and among the top 12 supply areas had

farmers from Masansa, Chisamba, Mkushi farm block, farmers in Kapiri Mposhi district

29

and Nkolonga. Most farmers in this area are concentrated in the top four deciles. About
19% of the tomatoes in Soweto market are from this area, and most of their tomatoes
were supplied in the high price months of February to April 2007 and December 2007 to

January 2008.

Eighteen percent of the tomatoes that enter Soweto market come through traders
(Channels 4-6). The tomatoes from the traders are originally from the farm areas but are
channeled through these intermediaries before they finally reach Soweto market. Channel
4 represents the tomatoes that come from the small farm areas to the traders, while
channel 5 represents tomatoes from the medium farm areas to the traders and finally

channel 6 representing tomatoes from the large farm areas to the traders.

Tomatoes from the different supply areas are then wholesaled in Soweto and Bauleni
markets and then eventually channeled out to the retail outlets. Among the various retail
outlets are the open air markets which account for 67% of the volumes of tomatoes,
followed by the Ka sector with 24%, with the remaining 9% being transacted in the
grocery mini marts (5%), large super market chains (<1%), large independent
supermarkets (<1%), and the remaining amount accounted for by gifts and private

household production and consumption (Table 2.5).

The informal retail system, in the form of the open air markets and the ka sector

dominates tomato retail. Almost 80 % of the FFV sales are carried out in the open air

30

market and the ka sector retail outlets, with only 5% share in the large supermarket chain

outlets and only 1% in the large independent super markets.

31

Table 2.7: Retail Outlet Market Shares on Overall Food (Lusaka)

 

Market Group [Retail Share for Share for Share for Share for Share for

 

Outlet all foods all FFV Vegetables fruits tomatoes

Open Air Market 0.32 0.55 0.64 0.45 0.67
Ka Sector 0.16 0.20 0.20 0.21 0.24
Grocer / Mini mart 0.21 0.01 0.02 0.003 0.05
Own Production 0.02 0.09 0.07 0.13 0.0]
Private HH 0.02 0.01 0.01 0.01 0.01
Gift 0.03 0.06 0.04 0.09 0.01
Large Independent

Supermarkets 0.01 0.01 0.01 0.01 0.01
Large Supermarket Chains 0.09 0.05 0.02 0.09 0.003
Butcher 0.14 0.00 0.00 0.00 0.00
Small Supermarkets 0.01 0.001 0.001 0.001 0.00
Other Purchasing Channel 0.01 0.00 0.00 0.00 0.00
Baker 0.001 0.00 0.00 0.00 0.00

 

Source: Food Security Research Project Urban Consumption Survey Data 2007

The broader literature12 on supermarket expansion in the developing world shows that the
general pattern of their development has mainly been through the spread of foreign direct
investment (FDI). Zambia is no exception. Much of the FDI in supermarkets in Zambia
is from South Africa where the supermarket share of the national food retail is 55%”.
The shares in South Africa are similar to those found in some Latin American countries
such as Argentina and Chile”. In Zambia however, the growth rate of these supermarkets
has not been as fast as in these parts of the world and hence the small share they have in

the retail outlet markets.

Further analysis on the retail outlet market shares for all FFV purchases made by the
households by the expenditure quartiles was conducted, and the results show that the
traditional retail still ranks highest among all the retail outlets used by all the expenditure

quartile groups (Table 2.6). In the two lowest income quartiles, the open air markets and

 

‘2 Reardon and Timmer, 2006; Tschirley 2007
'3 Weatherspoon and Reardon, 2003.
'4 Weatherspoon and Reardon, 2003.

32

the ka sectors combined have shares of over 90%, while the top two income quartiles (3
and 4) have shares of at least 80%. Households in the highest income quartile tend to use
the formal retail outlets (specifically the small supermarkets and the large supermarket

chains) more than the other income quartile groups.

Table 2.8: Retail Outlet Market Shares for all FFV Purchases by Income Quartile

 

Expenditure Expenditure Expenditure Expenditure

 

Market group/Retail outlet quartile 1 quartile 2 quartile 3 quartile 4

Open Air Market 0.67 0.70 0.62 0.53
Ka Sector 0.26 0.22 0.26 0.27
Grocer / Mini mart 0.002 0.002 0.009 0.037
Small Supermarkets 0.00 0.00 0.001 0.001
Large Independent supermarkets 0.00 0.00 0.00 0.01
Large Supermarket Chain 0.002 0.004 0.02 0.06
Butcher 0.00 0.00 0.00 0.00
Baker 0.00 0.00 0.00 0.00
Private household 0.01 0.02 0.02 0.02
Other Purchasing Channel 0.00 0.00 0.00 0.00
Own Production 0.02 0.03 0.05 0.06
Gift 0.02 0.02 0.02 0.02

 

Source: Food Security Research Project Urban Consumption Survey Data 2007

An examination of the retail outlets shares for tomatoes by expenditure quartiles, also
reveals that the open air markets and the ka sectors combined have the largest retail outlet
market share (Table 2.7). The highest income quartile has a combined retail outlet market
share of 85% in the open air markets and the ka sector while the other income quartiles
all have over 90% share. The highest income quartile are the main group that use the
grocery/mini mart and large independent supermarkets for the purchase of tomatoes with

shares of 7% and 1% respectively.

33

Table 2.9: Retail Outlet Market Shares for Tomato Purchases by Expenditure
Quartile

 

Expenditure Expenditure Expenditure Expenditure

 

Markeflup/Retail outlet quartile 1 quartile 2 quartile 3 quartile 4

Open Air Market 0.58 0.64 0.65 0.55
Ka Sector 0.37 0.28 0.30 0.30
Grocer / Mini mart 0.00 0.00 0.002 0.07
Small Super markets 0.00 0.00 0.00 0.00
Large Independent Super markets 0.00 0.00 0.00 0.01
Large Supermarket Chain 0.00 0.00 0.003 0.00
Butcher 0.00 0.00 0.00 0.00
Baker 0.00 0.00 0.00 0.00
Private households 0.03 0.03 0.03 0.04
Other Purchasing Channel 0.00 0.00 0.00 0.00
Own Production 0.01 0.03 0.01 0.01
Gifi 0.02 0.02 0.00 0.01

 

Source: Food Security Research Project Urban Consumption Survey Data 2007

Evidently, the informal sector comprising the open air markets and the ka sector are very
important. The formal sector has a very low percentage share for the transaction of FFV
and especially for tomato, despite the manner in which it is well organized and the
infrastructure in place. In view of the high percentage share of FFV transactions
occurring in the two identified informal channels, it would be paramount to ensure that
the performance of this sector is enhanced by way of identifying means through which

there would be a more efficient handling of the volumes of FFV that pass through it.

2.4.3 The ‘Modern’ Sector: Supermarkets and Processors
The modern sector of the tomato system is composed mainly of supermarkets and
processors. The supermarkets that dominate this sector are Shoprite, Melissa and Spar
while the processors include Freshpikt and Rivonia. The supermarkets and the processors

in this sector jointly have a 9% share in the tomato system. The following section gives

34

some details of these supermarkets and processors and also looks at the tomato

procurement system they have adopted.

i. Shoprite Supermarket/Freshmark
Freshmark serves as a wholesale procurement and distribution channel for tomatoes
supplied to all 17 Shoprite retail outlets countrywide. Shoprite is the largest super market
chain in Zambia and mainly relies on Freshmark for all its FF V requirements. It however

handles less than one percent of the tomatoes consumed in the country.

In its tomato procurement system, F reshmark currently has four farmers that supply it
with tomatoes; three commercial farmers and one small scale farmer. All of these farmers
are located in Lusaka province. The small farmer is located in Makeni, South of Lusaka
city. One of the large commercial farmers is located in Chisarnba area while the other two

are South of Lusaka city in Kafue area.

Ninety percent of the tomatoes supplied to Freshmark come from the three commercial
farmers and the remaining 10% comes from the small scale farmer. Ambrosia farm
accounts for 40% of the supply while the other two commercial farmers account for the

remaining 50% with each one supplying approximately 25%.

To qualify as a tomato supplier to F reshmark, the suppliers have to adhere to a number of
quality standards that are above what would be expected in an open air market. Some of
Freshmark’s quality requirements are, firm, champagne red color tomatoes, free of any

blemishes and able to have a shelf life of 3 days at the time of delivery.

35

Freshmark mostly prefers to have large farmers as tomato suppliers as they are more
reliable and stick to the terms and conditions of the contracts they enter into. The
procurement manager indicated that small farmers, other than having production
constraints which hinder them from supplying required quantities and their inability to
produce a product that meets Freshmark’s quality requirements, have a tendency to break
the contracts and supply a market that offers a better price at a given point in time. The
small farmer that currently supplies tomatoes to Freshmark is a very committed farmer

but has land area limitations that hinder him from expanding his tomato production.

In cases where the local tomato suppliers are not in a position to meet Freshmark’s
demand, Freshmark outsources tomatoes from Freshmark South Afiica. This is

particularly the case in the rainy season when the local tomato supplies are very low and

prices high.

An average of 4 mt of tomatoes is supplied to the various Shoprite retail outlets every
week. Seventy-five percent of these tomatoes end up in Lusaka while the remaining 25%
go to the Shoprite retail outlets outside of Lusaka. Tomatoes are very important in the
vegetable procurement system as they rank second from potatoes in sales volumes, and
rank fourth in Shoprite’s overall FFV supply system with bananas, potatoes and apples

taking the lead in this order.

Freshmark usually seeks to maintain stable prices during the course of the year. To

achieve this, the contracted farmers are offered less variable prices for their tomatoes for

36

the whole one year contract period they enter. Due to this pricing policy, during the peak
supply season when the tomato prices are generally lower Freshmark offers its farmers
higher prices than what the market is offering, and when the tomatoes are in short supply
and prices expected to be higher, Freshmark’s prices would be lower. It is during the high
market price period that most contracted farmers (particularly the small ones) would

default and sell their tomatoes where the price is higher.

ii. Melissa Supermarket
Melissa supermarket is a Zambian grocery store chain with three outlets in Lusaka city
located in Northmead, Kabulonga and Matero. The Matero outlet is the most recently

opened and forms the focus of the following discussion.

Among all the F FV products purchased by Melissa, tomatoes are important, however
onions top the list in importance. Melissa has an internal procurement system for
tomatoes with contractual arrangements with three commercial farmers, Eco Veg, Agir
Link and Lilayi farms. Each of these farmers supplies Melissa with an assortment of

FFV, but only Eco Veg supplies them with tomatoes.

In addition to procuring tomatoes from the commercial farmer, Melissa also obtains some
fiom small independent farmers. These independent farmers are basically walk-in
suppliers without contracts with Melissa, but meet the quality requirements for firm, semi
ripe, blemish- fi'ee tomatoes. Melissa has therefore adopted a dual procurement system

which enables it to cushion the effects of price fluctuations and unstable supplies.

37

With the dual procurement system that it has adopted to manage the supplies of tomatoes
from both sources, Melissa supermarket ensures that it has a weekly tomato supply of
350 kg. Eco Veg supplies them with tomatoes on Mondays, Wednesdays and Fridays

while the other suppliers supply the tomatoes on the other days.

Melissa supermarket has a fixed price arrangement with Eco Veg over each contract
period, which may vary fi'om a few months to one year. During the contract period,
irrespective of whether the market price of tomatoes drops or rises, Melissa supermarket
pays Eco Veg only the agreed amount. In periods when the supply of tomatoes on the
market is high and the market price lower than the negotiated price with Eco Veg,
Melissa procures most of its tomatoes from the other suppliers (small independent
farmers). On the other hand, when the supply of tomatoes on the market is low and the
market price is higher than the negotiated price with Eco Veg then Melissa procures most
of its tomatoes from Eco Veg. This procurement arrangement enables Melissa to keep its
prices fairly stable over a given period of time. With supplies from both sources, Melissa
averages out the prices received; given the fixed price fiom Eco Veg and the variable

price from the other suppliers which may be lower or higher than the Eco Veg price.

Melissa is comfortable with this dual supply system and does not have any preference for
either. The benefits of having such a system are better than having one supply source.
Considering the tomato supplies from the farmers, Melissa supermarket has preference
for supplies from large farmers as their quality of tomatoes is better than what is obtained

from the smaller farmers.

38

iii. Spar Supermarket
Spar supermarket is the newest supermarket chain in Zambia with its origins in the
Netherlands. The first Spar retail outlet was opened in 2004. It currently has six outlets
countrywide; two in the Southern province towns of Livingstone and Choma, and four in
Lusaka province: Downtown Spar, Soweto Spar, Arcades Spar and Chawama Spar which

was just recently opened in mid 2008.

Each of the Spar outlets is run as an independent operation by its own manager, and each
with its own FFV procurement system and pricing policy. Downtown Spar markets a
wide range of vegetables such as carrots, peppers, onions, cucumbers, tomatoes, potatoes,
green beans, and others. Most of the vegetables and other fresh produce they sell come
from three large farms: Buyabarnba farm, Osuma farm and Birchwood farm. The large
farms account for 60% of the vegetables they are supplied with while the remaining 40%
is supplied by small farmers and independent traders who deliver the tomatoes to their

premises.

On a weekly basis, downtown Spar sells an average of 125kg of tomatoes. To ensure that
they have a steady supply of tomatoes throughout the year, the store heavily relies on

Buyabarnba farm which consistently has tomatoes throughout the year.

iv. Freshpikt
Freshpikt is the dominant F FV processing firm in the country. At present, it produces its
own tomatoes and supplies its processed products to the grocery mini-marts, large

supermarket chains, the large independent supermarket retail outlets and exports 5%.

39

Compared to Soweto market which had an 83% share of raw tomatoes in the system,

Freshpikt had an 8% share for raw tomatoes in 2007.

Freshpikt produces 18 different canned products which include baked beans, mixed
beans, tomato puree, tomato paste, tomato and onion mix, whole peeled tomatoes and an
assortment of fruit chunks, jams and juices from pine apples. Tomatoes, beans, sweet
corn and onions are the main vegetables they process and tomatoes are the major

ingredient used in most of their salty canned products.

Freshpikt currently sources all of its tomatoes from its own 40 ha farm plot in Lusaka
East. At 50 mt per week, all year round, the plant is operating well below its capacity of
60 mt per day. It has plans to step up its processing volumes for tomato products once it

engages small tomato grower cooperatives on a contractual basis in its supply chain.

v. Rivonia
Rivonia is another FFV processing firm specialized in the production of tomato sauces.
They use local raw tomatoes and imported tomato paste for their sauces. They currently
procure 540 Kg of tomatoes per week from independent tomato growers in Lusaka
province. At present, the volumes of tomato that come fi'om the farm areas to Rivonia
have a share of less than 1% in the system. As with Freshpikt, its processed products end
up in grocery mini marts, large supermarket chains and the large independent

supermarket retail outlets.

4O

2.5 Price Behavior
2.5.1 Weekly Wholesale Prices in Soweto Market
Soweto market is the main wholesale market in Lusaka and serves as the main source of
tomatoes for most of the retail outlets in the city. The graph presented below shows the
weekly wholesale per kg tomato prices that prevailed in Soweto market over the period
Jan 2007 to July 2008. These are the prices received by farmers and traders selling in the

market.

Figure 2.3: Weekly Soweto Wholesale tomato prices January 2007 to July 2008

 

3000.00 -

2500.00

2000.00

1500.00

 

1 000 .00

Mean price Ikg

 

500.00

0.00 -

 

 

 

I I I I I I I I I I I I I I I I I I I I I I I I I I

3 9 15 21 27 33 40 46 52 5 11 17 23 26
WK WK WK WK WK WK WK WK WK WK WK WK WK WK
07 07 07 07 07 07 07 07 08 08 08 08 08 08

Week

Source: Food Security Research Project - Tomato price data 2007-2008

41

During this period, tomato prices were quite variable in Soweto market. It was observed
that, despite strong seasonal patterns, there is a fair amount of price variation within a
given season. A notable feature in the graph over the whole period is the sharp price
declines experienced in April 2007 (15 Wk 07), December 2007 (49 Wk 07), March 2008

(11 Wk 07) and June 2008 (25 Wk 08).

2.5.2 Weighted Average Prices by Marketing Channel
Going by the channels identified in the channel map (Figure 2.1), table 2.8 shows the
weighted average prices for a kg of tomatoes in each of these channels. Taking a look at
the channels for tomatoes that get into Soweto from the farm areas, it can be observed
that the farmers from large farm areas (channel 3) received the highest prices of ZMK
1,138 followed by farmers in the small farm areas (channel 1), ZMK 1,055 and finally

farmers in the medium farm areas (channel 2) receiving ZMK 1,007.

Interestingly, when we look at the channels for tomatoes that pass through the traders
before they reach Soweto market, we observe a similar price pattern. The tomatoes sold
by traders buying form the large farm areas (channel 6) are sold at the highest price,
ZMK 1,223, followed by those sold by traders buying from small farm areas (channel 4,
ZMK 989) and finally from traders buying from the medium farm areas (channel 5, ZMK

932).

42

Table 2.10: Weighted average tomato prices by market channel

 

 

Channel Weighted Avg. Price
Number Channel Description (ZMK)

6 Sales by traders buying from large farm areas 1,223

3 Direct farmer sales into Soweto from large farm areas 1,138

1 Direct farmer sales into Soweto from small farm areas 1,055

2 Direct farmer sales into Soweto from medium farm areas 1,007

4 Sales by traders buying from small farm areas 989

5 Sales by traders buying from medium farm areas 932

 

Source: Food Security Research Project - Tomato price data 2007-2008

2.5.3 Tomato Wholesale and Retail Prices
A comparison of tomato prices for Soweto market and four selected retail markets,
namely Spar, Shoprite and Melissa supermarkets, and Chilenje open air market, was
made by examining the price trend over the period between January 2007 and July 2008
(figure 2.4) We observe that Soweto wholesale prices for tomatoes for the whole period
averaged ZMK 1179, while the retail prices in the selected retail markets were ZMK
3,450 for Chilenje open air market, ZMK 3,545 for Melissa supermarket, ZMK3,408 for

Spar supermarket and ZMK 3,390 for Shoprite supermarket (table 2.9).

43

moo~-noo~ San Sta 89:0» I 8&9“. 5.83m Btsomm noon. ”850m

gm I
2.00 .8505 3:
3:59.54 .8830: I

 

 

 

0.520. I
02.33.02.038. . .
3x55.
is;
momomomowowowomowomomomowomonomonononouono nonono No homo no no RDNONONONO homouonohvho
v5>v$>v5>éthvgééééééééééééééévtSéEéééééééévfigééééééé
mmwmvmmmomwrwwv—Nwow m w v N mm Pmmvnvmvnv Smmnnmmmm wnmmhwmmnn rmmwhwmwnr E. m n m n
p p — — — h n - — h n P _ — _ p h — p h h p p p h — p P p p p b h n h h h F In b
1.00.0
.. a s .. a. .. . . coo-c.3333... ..s..... ..s
. .e .. ~ soc. . . \ID“ I ’ M IOO.OOON
d
u.
a
C
d
a
J
u
..ronXHx3M
M
l.\
loodoom

 

 

 

351— :33— 15 2322—3 2. wfiotm Snack. "fin enema

44

 

 

On the basis of these mean weekly prices observed for these markets (figure 2.4), we see
that Chilenje market followed a very similar price pattern as Soweto market. Much of the
tomatoes in Chilenje market are obtained from Soweto market and the prevailing prices
in Chilenje reflect a fairly stable price mark up averaging ZMK 2,284 per kilogram of
tomato. To demonstrate the fairly stable price margin over the period, the price margin
was graphed (figure 2.5). However, in mid March and late June (Week 13 and 25
respectively) there were price margin spikes experienced in Chilenje market. In these
periods, Soweto market . experienced some price drops and despite these price drops,

Chilenje market seems to have maintained their price mark ups thereby resulting in the

high price margin.

Melissa supermarket maintained a fairly stable price over the period with the tomato
price averaging ZMK3,545 per kilogram. Shoprite on the other hand seemed to follow

the traditional retail market (Chilenje) prices in a stepwise fashion.

Of all four retail markets evaluated, Spar supermarket had the most stable year round
prices for tomato at a mean price of ZMK 3,400 for most of the year. In both the peak
and low supply periods, it maintained this stable price with the exception of the low

supply period of January when it had a low price of ZMK 2,700.

45

Table 2.11: Mean Tomato Prices for Wholesale and Retail Outlets in Lusaka

(January 2007 to July 2008)

 

 

Outlet Type of outlet Mean tomato price
(ZMK)

Soweto Wholesale 1 179

Shoprite supermarket Retail 3390

Spar supermarket Retail 3408

Chilenje open air market Retail 3450

Melissa supermarket Retail 3545

 

Source: Food Security Research Project — Tomato price data 2007/2008

46

3on—

wo mo 00 00 mo 00 00 mo 00 NO NO NO NO ho ho ho NO NO NO NO ho ho ho no no NO
v5)¥>>¥>>¥>>¥>>¥>>¥>>¥>>¥>>¥>>¥>>v=$v<$¥>>v<$¥>>¥>>¥>>v=$¥>>¥>>v5>¥>>¥>>¥>>v5>
0N 0N ON 5 r v r r P D m N Nm 0? 0? av 0? kn mum on 5N VN FN 0 r m r N r m 0 m

_ _ - _ _ _ P — _ — _ L — P — — - — - - _ — - _ _ _

 

 

 

 

 

 

I 8.08 F
Ioodoom
0
N
m
m.
6
I868» .m
m.
a
w
m
m.
Ioodoov u
)
7.
m
x
5
(
10988
I888

 

 

ace—.82 =33— omaoaau ..8 Swan: 03...— "flu 0.5»:—

47

2.5 Summary and Conclusions
2.5.1 Importance of Tomatoes
Fresh fruits and vegetables are a major food item purchased by households in Lusaka,
Kitwe, Mansa and Kasama urban centers. In all these cities, the budget shares for all food
items purchased by the households shows that FFV are an important food item as they
rank third after cereals/staples and meat/eggs respectively. Detailed examination of the
specific FFV items consumed by the households reveals that tomatoes rank first in the
FFV budget shares, with almost 10% of the expenditure on all FFVs going to tomatoes.
Further examination of the budget shares for tomatoes over all F FV by expenditure
quartiles in Lusaka also reveals that tomatoes rank high in expenditures taking up an

average of 10% of the F FV budget for households.

Based on these results, it is clear that tomatoes are an important F FV item and that takes
up a substantial amount of the consumers’ incomes, thereby impacting on the

households’ purchasing power.

2.5.2 The Tomato Subsector
The tomato subsector in Lusaka is made up of tomato farmers, traders, wholesalers,
processors/assemblers, and retailers. The farmers supplying tomatoes in the system are
from large, medium and small farm areas, however, the large and medium farmers

dominate the system and supply about two thirds of the tomatoes in the system.

Among the wholesalers, processors/assemblers and retailers in the tomato subsector,

these actors make up the traditional sector and the modern sector of the tomato subsector.

48

The traditional sector refers to the informal sector which is mainly made up of the
Soweto and Bauleni markets, and accounts for over 90% share of tomato volumes at
wholesale level. At retail level, we have the open air markets and the ka sector which
collectively account for 91% of the retail sector. The modern sector, on the other hand is
a formalized sector of the tomato subsector. It is comprised of the FFV processors,
Freshpikt and Rivonia, and Freshmark, a large wholesale operator. At retail level, the
modern sector is made up of the large independent supermarkets, the large supermarket

chains, mini marts and small supermarkets.

Atleast Sixty six percent of all tomatoes entering Soweto market are directly marketed by
farmers, while 18% are marketed through traders. Traders buy the tomatoes either at the
farm gate, then transport and sell them in Soweto, or from the farmers at Soweto market
then sell them to wholesalers there. Eight percent of the tomatoes from the farm areas
were sold directly to the wholesalers in Bauleni market, and less than 1% was sold to
Freshmark wholesalers and Rivona. The remaining tomatoes in the system are grown by

one of the F FV processing firms, Freshpikt.

Farmers from twelve main geographic areas dominate the system and account for 68% of
all the tomatoes in the system; the other 32% was split among over 150 other supply
areas. These farmers supplied tomatoes to Soweto market at different times of the year
with some of them predominately supplying them in the low price months while others

supplied them in the high price months.

49

In the period January 2007 to June 2008, tomato prices were quite variable, and some of
this variability in the prices was the normal price variability one would expect due to
seasonality of production. Notable high price months for tomatoes over this period were
in February to March 2007, September and November 2007, and December 2007 to
February 2008. The observed low price months were around April to August 2007,
November 2007 and March 2008. Also worth noting were the sudden price collapses that

were experienced in April 2007 and March 2008.

Analysis of the farmers that supplied tomatoes in the market reveals that among the top
twelve supply areas, the areas Masansa, Choona, Lusaka West and Manyika are partly
responsible for the price collapse experienced in April 2007. These four areas supplied
59% share of tomato volumes during this period. In the case of the price collapse that
occurred in March 2008, farmers from Choona and Masansa may have caused it as they

account for 49% of the tomatoes in the market at that time.

Once the tomatoes from the different supply areas arrive in the wholesale markets, they
are then channeled out to the consumers through the various retail outlets. The traditional
retail sector, comprising the open air markets and the ka sector dominate the retail
market, with 91% share. Analysis of the UCS also shows that the traditional retail sector
dominates with over 90% of FFV sales occurring in it. Further examination of the retail
outlets used for F FV purchases by income quartile groups also shows that the traditional
retail sectors dominates with very few purchases being made in the modern sector

supermarkets.

50

The modern sector is mainly made up of supermarkets and processors which jointly have
a 9% share in the tomato system. Shoprite, Spar and Melissa are the dominant

supermarkets while Freshpikt and Rivonia are the main F FV processors.

The analysis conducted in this chapter has shown that tomatoes are an important F F V
item among urban consumers in Zambia. It has also further shown us the dominance of
both the wholesale and retail traditional sectors of the tomato subsector. Given the
dominance of the traditional wholesale and retail sector in the tomato sub sector, and the
poor infrastructure that exists, particularly in Soweto wholesale market, it is important
that particular attention be paid to them so that they are better able to serve the needs of
both the sellers and the consumers. Some of the key areas that need improvement for the
better function of these systems are in the improvement of market infrastructure (roads,
physical buildings, sanitation, and drainages), market information and cold chains.
Therefore, developing these markets has the potential to increase the incomes of farmers
due to the efficiency that would result from it. In addition to this, their improvement
would pave way for further upgrading the systems to standards that are comparable to the

modern sector of the tomato subsector.

51

CHAPTER 3
TOMATO PRICE VARIABILITY AT WHOLESALE LEVEL: COMPARING
SOWETO MARKET (ZAMBIA) WITH OTHER WHOLESALE MARKETS
ACROSS THE WORLD
This chapter examines tomato price variability for wholesale prices in Soweto market,
Zambia and compares it with variability of other tomato wholesale prices across the
world: United States of America (Chicago), Taiwan (Taipei), Costa Rica (San José), and
Sri Lanka (Colombo). United States of America (USA), Taiwan, Costa Rica and Sri
Lanka were chosen for comparison with Zambia because of the wide range of levels of
market development in these countries, with USA and Taiwan being the most developed,

Sri Lanka expected to be similar to Zambia, and Costa Rica expected to lie somewhere

between these extremes.

We first discuss factors that influence price variability and predictability, followed by a
detailed presentation of the methods used in the analysis. Then finally, the results and

discussion of the analysis shall be presented.

3.1 Factors Influencing Price Variability and Predictability

Price variability refers to the state of prices being variable over a given period of time,
while price predictability on the other hand refers to the degree to which prices can be
forecasted correctly. In general, the higher the price variability for a given product, the
more difficult it is to predict the price for that product. Over the course of a year, the
prices of a product can be fairly variable due to the seasonal production of the product.
This kind of price variability is expected to Show some consistency from year to year.

However, because the precise seasonality of production can vary from year to year due to

52

variable weather patterns, the seasonal pattern of prices is not fully predictable. Since
product prices directly affect the incomes that a farmer makes, an improved knowledge of
the patterns of price variability and the forces behind it might help them better understand

and manage their price risks.

The variability of prices and the degree to which prices can be predicted is influenced by
a number of factors. Many of these factors have to do with supply conditions for a given
product, such as the seasonality of supply and supply shocks that the product could be
subject to. A third factor has to do with random day-to-day variations in the quantity of
product that arrives in the market; perishable horticultural products are especially
vulnerable to this type of variation. Finally, improved grades and standards can improve

price predictability for a farmer without affecting price variability.

i. Seasonality of Supply
Seasonality refers to fluctuations in product output related to the season of the year.
Agricultural products, whose production is affected by weather patterns over the course
of the year, are usually subject to seasonality of supply. Zambia is warm all year round
but has three distinct seasons”. Between December and April the weather is hot and wet;
from May to August it is cooler and dry; between September and November conditions
are hot and dry. Average high temperatures during the hot wet and hot dry seasons range
between 77°F to 95°F (25°C to 35°C), while in cool dry season the variation increases

ranging from 43°F to 75°F (6°C to 24°C).

 

'5 Information on the seasons and climate in Zambia is drawn from

http://wwwwordtravels.com/Travelguide/Countries/Zambia/Climate/

53

In the hot wet season, disease prevalence, pest and weed infestation in vegetable crops
are high. Crop management requirements for diseases, pests and weeds are therefore high
during this season and as such, the amount of vegetable production that takes place is
limited. As a result of this, there is an overall short supply of flesh vegetables in the
market during this season. In the cooler dry and the hot dry seasons, disease prevalence,
pest and weed infestation are not as pronounced as in the hot wet season, and as a result,
the cost of managing a crop during these two dry seasons are lower. Due to the much
more favorable vegetable growing conditions in these seasons, particularly the hot dry
season, the supply of flesh vegetables on the market is higher in these two seasons. These
seasons are however faced with higher irrigation costs as they do not depend on rainfall
for irrigation, but we expect that the cost of irrigation will be lower than the cost of pest,

disease and weed control in the wet season.

Seasonal climate patterns in Zambia therefore greatly influences seasonality of
production and supply of vegetable crops and other crops alike. Other factors that could
affect seasonality of supply and ultimately also influence price variability and
predictability include the degree of integration of product markets, the extent of irrigation

and, more generally, the ability of a farmer to control their production environment.

a. Integrated Product Markets
Integrated markets may be considered as an interconnection of several markets not
located in the same geographical area. Markets are interconnected by virtue of the
common products they buy and sell and the movement of products between these markets

based on the supply and demand conditions in each market. The end result of integrated

54

markets is mainly in the provision of better signals for optimal production and
consumption decisions and subsequent pricing efficiency. Well integrated markets
therefore improve security of supply of a product and ensure that an equilibrium point is
reached in that product market. Such an equilibrium is achieved when the flow of a

product is from high supply areas to low supply areas.

Consider the case of two markets located in different production/consumption zones,
which have different seasonal patterns of production. One market produces and sells the
product for the first half of the year, while the other market produces and sells it in the
other half of the year. If there is no trade between the two markets, each will have large
price fluctuations over the course of a year. In the case where there is trade between
them, thereby promoting integrated markets, price seasonality in each would be greatly

reduced.

Despite the reduced price variability that could accrue from having integrated markets,
not all markets are integrated. Some the factors that inhibit market integration include
high costs of transporting products from one area to another, the absence of cold chain

facilities and limited relevant market information.

High costs of transportation hinder integration of markets by impeding the transporting of
products from high supply areas to low supply areas. The high transport cost could be
manifest in the form of high fuel costs, long distance between markets, an inadequate

road network or poor condition of the roads. To the extent that seasonal production

55

patterns differ across markets, reduced trade due to high transport costs results in higher

seasonality of supply in each area.

Cold chain systems enable the transportation of perishable products over longer
distances. The integration of markets can be aided by the presence of cold chains. The
lack of cold chains means that markets will be integrated only over smaller geographic
areas. Where there is a cold chain in place, to the extent that seasonal production patterns

are different across markets, this would reduce seasonality in all markets.

In the presence of market information, a farmer in a high supply area can make an
informed decision about taking their product to an area where the supply is low. The
effect of this would be to lower prices of the product in that area. In the absence of
market information, suppliers could possibly end up taking their product to an area where
the supply is high and would further depress the price of the product in that area.
Therefore, poor market information limits the possibility of market integration and

subsequently seasonality of supply would remain a prevailing concern.

b. Irrigation/Ability to Control Production Environment
Seasonality of supply is often affected by limited water supplies or poor production
environment. Considering the case of limited water supply for crop production, a farmer
could mitigate supply effects resulting from this by irrigating their crop. In the case of a
poor production environment such as suboptimal temperatures and high humidity, or
disease and pest infestation, farmers can avert supply effects from such by controlling

their crop production environment through the use of green houses, insecticides and

56

fimgicides. If a farmer has access to irrigation and other technology that enable them
control their production environment, seasonality of supply for a particular crop could be

greatly reduced.

ii. Supply shocks (disease or pest outbreak, drought, flood)
A supply shock is an event that suddenly increases or decreases the output of a product or
service temporarily. The result of this sudden change in supply changes the equilibrium
price of the product or service. A negative supply shock (a sudden decrease in supply),
will cause a rise in the price of a product or service while a positive supply shock (a
sudden supply increase) will lower the price of a product or service. Some of the
common supply shocks that would affect the supply of an agricultural product include
disease or pest out breaks, drought and flood. Alternatively, especially good weather

could lead to unexpectedly high supply and low prices.

In the absence of mitigation measures, supply shocks could be accentuated, and
subsequently have adverse effects on agricultural production and the supply of the
agricultural products. Some of factors that could help in mitigating the possible effects of
a supply shock include the use of irrigation, the use of a controlled production
environment (e. g. greenhouses), access to pest and disease control inputs and farmer

knowledge of how to control pest and disease inputs.

Irrigation and control of production environment: Supply shocks that could result
fi'om adverse weather conditions such as drought or flood could be avoided through the

use of irrigation or the use of green houses that have a well regulated water supply. In the

57

case of a flood, its effects could also be avoided through the use of a controlled

production environment such as a greenhouse.

Access to pest and disease control inputs: Easy accessibility to chemical pest control
inputs reduces susceptibility to a pest or disease outbreak. However, the accessibility of
these inputs is subject to the general development of the input markets in a country and
also the credit or cash availability to the farmers that use these inputs. Poorly developed
input markets and the financial limitation of farmers would mean that they would not be

able to counter the effects of a disease or pest outbreak on their agricultural product.

Farmer knowledge: If on the other hand a farmer has easy access to pest and disease
control inputs but lacks the knowledge on how to properly use them, then the farmer
would not be able to either identify the disease or pest problem, or to use the correct
control inputs, or to administer them incorrectly. The problem may further be accentuated
by the absence of extension services in their area and the absence of an early warning

system against pest or disease problems moving into the area.

iii.Random Fluctuations in Quantity Supplied to the Market
Already discussed is the issue of seasonality of supply and that of supply shocks and how
they tend to cause price variability. Another factor that could influence price variability is
the random fluctuations in the tomato quantities supplied to the market. Random
fluctuations of the quantity of a product supplied in a particular market may be by the day
' or by the week. In both case, such fluctuations would entail that the price for the produce

would be variable as would be dictated by the supply and demand situation in the market.

58

For any given FFV, random fluctuations in the quantities supplied to a market may be the
result of the presence of a varying number of suppliers in the market at different times of
the day or days of the week, uncoordinated production and supply of the product in the
market, the absence of market information on the demand and supply conditions of a
product or the differences in marketing strategies (such as when to harvest and take the

produce to the market) adopted by the producers.

Therefore, even without a supply shocks or production seasonality, the quantities of
tomatoes that arrive at a market will show a random component fiom day to day or week

to week. The end result of this would be big effects in price variability and predictability.

iv.Grades and Standards
The factors discussed above influence both price variability and predictability. Some
factors are however specific to price predictability and these include grades and
standards. Grades and standards allow trading of a product on the basis of specific
parameters identifying their quality and other characteristics, thereby making the market
more transparent and reducing unpredictable variation in prices without necessarily
making prices less variable. Where there are poor or no grades and standards, a farmer
will not be certain of the price they will receive within a given range of prices being paid
at any one point in time. The use of more grades and more precise specification of those

grades increase price predictability for a given level of price variability of a product.

59

3.2 Hypothesis Testing

The level of price variability for a given product in a given market is related to the level
of development of the economy in which the market operates. In this context, a well
developed market is a market which (among other things) is capable of moderating the
effects of seasonality of supply and supply shocks and thereby experiences less price
variability. In more developed markets, better market information can reduce random
variability in quantities of a product arriving on the market as it would give an indication
of the supply and demand situation for a given product in different markets thereby
enabling producers of the product to channel that product to an appropriate market; better
information also gives sellers in a market more ability to plan the supplies they bring to
the market and to source those supplies from the most competitive market. With better
market information, suppliers of a given product would be knowledgeable about markets
that are in short supply of the product they are trading in. Based on this knowledge and
the coordinated efforts of several suppliers of the product, then the problem of random

variability in quantities of the product arriving in the market would be less pronounced.

More developed countries also often have stronger grades and standards that define the
prices at which a product would be sold. With grades and standards in place, then it
would be possible to make some price predictions with some degree of accuracy.
Therefore, in view of better market information, and grades and standards present in the
markets of better (more) developed economies, these markets are likely to have less price

variability and better price predictability than markets in less developed economies.

60

As a proxy for the level of economic and market development, per capita GDP (Gross
Domestic Product) in purchasing power parity (PPP) terms was used for Zambia and the
four other selected countries whose tomato price variability was analyzed (table 3.1).

Table 3.1: GDP Figures for Zambia and Other Selected Countries (Purchasing
Power Parity Terms)

 

 

Country PPP GDP
USA 45, 790
Taiwan 30, 126
Costa Rica 8, 295

Sri Lanka 4, 259
Zambia 1, 359

 

Source: World Bank, 2007
The hypothesis to be tested is that countries with higher PPP GDP (and thus with more
developed fresh produce markets) have less price variability and better price
predictability than those with lower GDP.
3.3 Data and Methods

3.3.1 Data
The data used in this analysis is tomato price data from Zambia and the four other
selected countries (table 3.2). Some of the price data is for periods as long as 83 months
(Taiwan), while some of it is for shorter periods such as 19 months as is the case with
Zambia. For specific details on the tomato price data on Zambia, please refer to section

2.1 .2 of this paper.

61

 

832a 5.38m 3:33 coon. ”mBEmN . no.m_mmmc._=.m

L .330DJ. _L_: It.
I. .C . I. r

 

_ .4 d 1.4.

 

 

.ngumxcmg Emu mm.>t OH mmr> >>u.>o .mwdehm

 

{#858835 sou aggregate“. Seaimtoesce he...

-.I._.:.
4 . .~ 444.

”<3 “mooQBoN .28 core Bees

“ tzucmEmhLud .95 .BSBHSE Smouuwoezom

 

momenta.» .«o omen.—

 

 

 

 

 

 

 

 

 

 

 

 

 

85 a é ace» seem as
can 55% E eeeaeeeome a 88 5:83 as»
ES? 28 02 cheese. 830 3585 ofiom £258 3 -Ev xooEmoEu m Aouosomv 8334 «ES
Sou H3580
55> :8 bone, 9 88 cease
£20 28 02 SEE 896 3 aouemeeouoha 3:88 3. >35 38200 835 Cm
30.5.5
5on obduE oven» was berg Aeoom 55800
new mexmofim £28 .37. .Ewto 9 coon 553 new a. cam
Ego 28 :5 80: 08:0 .3 83388me 3:88 am mansions 3an 09830 35m @815
a 8Q 88 H"5532
Bane Boo :3 mo 80389 can can one .830 9 ooom .9353 .3282
a 26: 8 box: .0922 08 80:0 3 somumueouoba 3:58 mm 953288 ban 5&8. 538,—.
.3 go 28
=& pogo—9% Ben
~83 8e ”Lou—38 woven» Room 59800
ceases 5 ocean emcee €96 85 a 88 5383 £45
09206 Eoo ofiom “mafia 05 80:0 .3 2.23355me 3:58 mm .36 xooahoau m 32. new SE 880
a
E can: .5 no
cauaouoEA—Eutn caueobggotn
5.3... 2.5 ..e 325 we anam— ceton 25H zones—3.5 oEaz ace—.32 €550

 

 

baa—33> cum...— 3an.—. .«o .

 

and: a... a can: can he 5358: "3 use...

62

From the table above, we observe that the frequency of the price data ranges from three
days in a week to daily prices. The Zambia price data has observations for only Monday,
Wednesday and Friday. Therefore, Monday, Wednesday, and Friday prices were selected

from all other countries so that analysis would be done on data with the same frequency.

3.3.2 Methods
The methods to be used in the analysis of the tomato price variability across the selected
countries are the analysis of the coefficient of variation and the conditional variance
analysis. The coefficient of variation is the simplest unconditional measure of price

variability while the conditional variance analysis is a measure of price predictability.

A high conditional variance implies that price predictability is low and vice versa for a
low one. In the case of the coefficient of variation, a low coefficient of variance indicates

low price variability and a high one indicates high price variability.

Coefficient of Variation

The coefficient of variation is a common statistic used for measuring the variability of
data. It is an expression of the dispersion of the observed data values as a percent of the
mean. It is a unit fi'ee statistic and therefore facilitates comparison of price changes in
different directions, across different time periods, different commodities, different

countries and currencies.

The coefficient of variation was calculated as follows;

63

 

 

Coefficient of Variation:

Where;

0- - the standard deviation for tomato prices.

P - the mean price for tomatoes.

Pt - the observed tomato prices.

Conditional Variance
The conditional variance is the tool of analysis that was used in determining the level of
tomato price predictability in the selected countries. To calculate the conditional

variance, the following steps were followed;
Step one — Generation of a prediction model

In calculating the conditional variance, a price prediction model had to be generated. The
prediction model used was based on a simple farmer price expectations process and not a
structural model. The model was a basic regression model which takes the following

form;

Pt=flo+fl1Xlt+ ........... +fl12Xllt+fl13Ii—l+fl147i+ut

Where;
Pt is the dependent variable and represents the predicted price for tomato in time t;

64

X it - are dummy variables for the months of January through December, excluding the

month which has a price closest to the mean. These dummy variables are included in the

model to take account of the influence of seasonality in production on tomato prices.

Pt—l - is the single period lagged price for tomato. This is included in the model to take

into account the influence of the previous prices on the current price and also because it is
the price a farmer will most likely look at in forming a price expectation.
T t - is a time variable in days. This is included in the model since it has an influence on

price predictability. This variable actually controls for seasonal price fluctuations.

For the full regression results containing the model summary and coefficients, please

refer to appendix 2.

Step two — Computation of the Conditional Variance from the regression outputs in step

one.

Using the residuals from the regression outputs, the conditional variance was calculated

using the following formula:

til Pt—Pt tz" “t

t=1 Pt t=l

 

 

Conditional Variance =

65

Where;

P - the observed tomato rices in the market,
1‘ P

1: - the redicted tomato rice in time t,
t P P

u t - the error term or residual, and

n — the number of price observations

The standardized residual is squared. Squaring of the residual therefore widens gap
between a big price prediction error and a small one. To ensure that the conditional
variance is unit free and comparable across time periods and countries, it is standardized

by first dividing the residual (u t ) by the price.

Based on the regression outputs, appendix 3 presents a plot of the residuals for tomato
prices for each country which provide a basis for comparing the extent to which these

country’s experiences positive and negative price prediction errors.

3.4 Results

3.4.1 Variability and Predictability of Prices
Computation of the yearly and mean coefficients of variation of nominal tomato prices in
Lusaka’s Soweto market and in other countries was conducted and later analyzed (table
3.3). Two points stand out: the difference in the mean coefficient of variation across all

countries and the difference in price variability by year for each country.

66

Table 3.3: Yearly and Mean Coefficient of Variation of Nominal Tomato Prices in

 

 

Selected Countries

Year/Country USA Taiwan Costa Rica Sri Lanka Zambia
2000 0.11 0.18 0.22 - -
2001 0.18 0.21 0.21 - -
2002 0.15 0.26 0.24 - -
2003 0.12 0.17 0.20 - -
2004 0.21 0.22 0.21 0.28 -
2005 0.19 0.18 0.20 0.21 -
2006 0.20 0.19 0.22 0.27 -
2007 - 0.16 0.21 0.24 0.24
2008 - - - - 0.26
Mean 0.16 0.20 0.22 0.25 0.25

 

Source: Costa Rica: www. pima. go. or; Taiwan. http: //amis. afa. gov tw/v-asp/ton v. asp: Sri Lanka:
www.sgg s.lirneasia. org, Zambia: Food Security Research Project tomato price data, 2007/2008; USA.

 

e&rirn=Run&type=termPrice&locChoose=location&commodigclass= a—llwithoutornamental

A closer look at the means across all the countries shows that all but Zambia and Sri
Lanka have different mean coefficients of variation. On the basis of the PPP GDP16
which was used as a proxy indicator for economic and market development, it is noted
that the USA which has the highest PPP GDP is the most developed of the five countries.
Examination of its coefficient of variation confirms this as it is the lowest. Zambia and
Sri Lanka on the other hand, with the lowest PPP DGP figures are expected to have the
least developed horticulture markets, have the highest coefficients of variation at 25%.
Taiwan and Costa Rica which have higher PPP GDP figures compared to Zambia and Sri
Lanka, are expected to have better developed horticulture markets, and their lower
coefficients of variation of 0.20 and 0.22 respectively, confirm this. A comparison of
Taiwan and Costa Rica shows that Taiwan, with a lower coefficient of variation than

Costa Rica, has a higher PPP GDP.

 

‘6 Reference to table 3.1.

67

A look at price variability in each country during individual years shows that the price
variability in the USA is consistently lower than all countries each year. In fact, the USA
never in any year reaches even the mean level seen in Zambia and Sri Lanka, while
Taiwan reaches those levels only once. From these results, we see that the most
developed horticulture markets (as proxied by PPP GDP), USA and Taiwan, consistently
show less variability than the two least developed horticulture markets of Zambia and Sri

Lanka.

The conditional variance for Zambia and the four other selected countries was also
computed and analyzed (table 3.4). From these results, one point that clearly stands out
is how the conditional variance figures for all countries fluctuate substantially from year
to year. We also note that the yearly conditional variance figures for the USA are
consistently much smaller than all other countries.

Table 3.4: Yearly and Mean Conditional Variance of Nominal Tomato Prices in
Selected Countries

 

 

Year/Country USA Taiwan Costa Rica Sri Lanka Zambia
2000 53 285 723 - -
2001 142 336 568 - -
2002 85 434 561 - -
2003 91 328 477 - -
2004 196 385 446 1252 -
2005 207 291 513 362 -
2006 1 l 1 310 459 896 -
2007 - 242 400 376 702
2008 - - - - 787
Mean 127 329 521 734 731

Source: Costa Rica: www. pima. go. or; Taiwan: hztgp //amis. afa .gov. tw/v-asp/ton-vw a 4;, Sri Lanka:
wwwflggs lirneasia. org, Zambia: Food Security Research Project tomato price data, 2007/2008; USA.

 

e&run=Run&type=termPrice&locChoose=location&commodityclass=allwithoutomamental

68

Zambia and Sri Lanka have the highest mean conditional variance and they are expected
to have the least developed markets of all five countries. From the PPP GDP proxy
indicator for economic and market development, the high conditional variance figures are
consistent with the low PPP GDP figures, indicating that the horticultural markets in
these countries are not that well developed and subsequently experience high price

variability.

Followed by Zambia and Sri Lanka is Costa Rica with a lower conditional variance of
521. Again, as proxied by the low PPP GDP, Costa Rica is expected to have a less
developed horticulture market. However, compared to Zambia and Sri Lanka, Costa Rica

has a market that is better developed.

Taiwan has a much lower conditional variance and a higher PPP GDP. As proxied by the
PPP GDP, Taiwan has a well developed horticulture market when compared to Zambia,
Sri Lanka and Costa Rica. A look at the low US conditional variance figures and the high
PPP GDP proxy for economic and market development, these results reveal that the US
horticulture market is the most developed one of the five countries as it has the highest

PPP GDP and the least conditional variance.

In the analysis of the conditional variance we observe that the ranking of the PPP GDP is
consistent with the ranking of the mean conditional variance for these countries. The
countries with well developed horticulture markets, as proxied by PPP GDP, have a

lower mean conditional variance than those with less developed horticulture markets.

69

For further comparison of the mean conditional variances for Zambia and the four other
selected countries, the mean conditional variance figures for each country were plotted
(figure 3.1). The higher the conditional variance, the less developed a country’s
horticulture market is as proxied by the PPP GDP indicator for economic and market

development.

Figure 3.1: Mean Conditional Variance for Zambia and Four Selected Countries

 

 

800
3
,._. 700
.2
5 600
2 500
(I
§ 400
'3 300
C
8 200
5 100
O
5 0

USA Taiwan Costa Rica Sri Lanka Zambia
Country

 

 

 

Source: Costa Rica: www. pima. go. or; Taiwan: hm //amis afa. gov. tw/v—asp/top-v. asp_; Sri Lanka:
www..ggs lirneasia. org; Zambia: Food Security Research Project tomato price data, 2007/2008; USA:
h J/markemews. usda. ov/ ortal/fv? af dm=full&dr=l& f ear id=1200002&r T e=wiz&s 2=tru

n o L o 1 {NL I L' o J". I II 1 L A 1
“an “Alt-1‘. II rn! nu \

       

3.4.2 The Problem of Predicting Sharp Price Declines
In fi'esh produce markets, the absence of cold chain facilities and the need for the product
to clear in the market can lead to sudden sharp price declines. The effect of this is in the
greater difficulty in predicting price drops compared to price rises for a given fresh
produce. To examine this matter, the mean absolute values of positive and negative

tomato price forecast error and the ratio of the mean negative price forecast error to the

70

positive tomato price forecast error for Zambia and the other selected countries were

computed and compared (table 3.5),

Table 3.5: Mean Absolute Values of Positive and Negative Tomato Price Forecast

 

 

Errors
USA Taipei Costa Rica Sri Lanka Zambia
Mean Absolute Value
Positive errors 0.0675 0.1120 0.1297 0.1294 0.1382
Negative errors 0.0746 0.1597 0.2022 0.2043 0.2443
R3“ °f “game and 1.1 1.4 1.5 1.6 1.7

positive errors

Source. Costa Rica: www. pima. go. or; Taiwan. t:tp //amis. afa. gov. tw/v-asp/top-v. asp; Sri Lanka:
www..ggs lirneasia. org, Zambia: Food Security Research Project tomato price data, 2007/2008; USA.

 

e&run=Run&tvne=terrnPrice&locChoose=location&commoditvclass=allwithoutornamental

A price prediction error is defined by the difference between the predicted price and the
actual price. The mean positive errors represent the mean of all the prediction errors
when actual prices were higher than predicted, and the mean negative errors represent the
absolute mean of all the prediction errors when actual price was lower than predicted.
Where the value for the mean of the (absolute value of) negative errors is higher than the
mean of the positive errors, this implies that operators in the market under consideration

have greater difficultypredicting price drops than they do price rises.

We observe that the US has the least ratio followed by Taiwan, Costa Rica, Sri Lanka and
Zambia. A comparison of these results with the PPP GDP proxy for economic and market
development of a country, we further observe that as this ratio increases, the PPP GDP
also decreases (figure 3.2). The conclusion that is drawn from this is that countries with
higher ratios have a problem of unanticipated sharp declines in tomato prices and hence

have poorly developed horticulture markets as proxied by the PPP GDP.

71

Figure 3.2: Comparison of the Ratio of the Absolute Mean Negative Errors to the
Positive Errors and the PPP GDP by Selected Countries

 

    

 

 

 

1.8 l I 50000

8 1.6 + T 45000 g

2 1 8

2 14 1 40000 E
a - l g

1» ~ 35000

g 2 1.2l 1 g
8’ 2 i 4 30000 (D

c 1: :1
a 11: J» 25000 a 'U
o > n. 2
E 2 0 8 .. n.
3 8 i 20000 ‘3’
:1 a o
3 0.6 n.
3 15000 a:
" o 4 l rc-

.5 . ‘ 10000 g
a 1 e
E 0.2 5000 E

o 2 .5 . lo
USA Taiwan Costa Rica Sri Lanka Zambia
L- Ratio of mean nwatIBEsREsgwjmm +Purchasing ’Po:wer Parity Gross Domestic] Product 1

 

 

 

 

 

 

 

Source: Costa Rica: www.pima.go.cr; Taiwan: hmzllamisafa.gov.tw/v-§p/top-v.a§p; Sri Lanka:
www.ggs.limeasia.org; Zambia: Food Security Research Project tomato price data, 2007/2008; USA:
h ://marketnews. usda. ov/ ortal/fv? af dm=full&dr=l& af ear id=1200002&re T e=wiz&s 2=tru
e&run=“ °tvne=tennPncelenoFL ‘ " n J r ‘ “ ‘ ‘

     

 

3.5 Summary and Discussion

From the results of the coefficient of variation, conditional variance analysis and the ratio
of the mean absolute values of negative to positive errors (RNPE), and with reference to
the PPP GDP which was used as a proxy indicator for economic and market
development, we see a clear consistent pattem that shows that tomato price variability is
higher and predictability is lower in countries that are considered to have horticulture
markets that are not very well developed. The results all point to the fact that countries
with well developed horticulture markets, as proxied by the PPP GDP, have lower

coefficients of variation, conditional variance, and RNPE.

72

The inverse relationship between the results of the conditional variance analysis and the
PPP GDP was visually compared by plotting the two (figure 3.3). A higher PPP GDP
corresponds to a lower coefficient of variation and vice versa for a lower one.

Figure 3.3: Comparison of the Coefficient of Variation and PPP GDP by Selected
Countries

 

Cooffeclont of Variation
Purchaaslng Power Parity Gross
Domestic Product

 

USA Taiwan Costa Rica Sri Lanka Zambia

E- Coeffecient of Variation {Purchasrng Power Parity Gross Domestrc Product 1

 

 

 

 

 

Source: Costa Rica: www.pima. go or; Taiwan: hm: //amis. afa. gov. tw/v—asp/top—v. asp; Sri Lanka:
www..ggs lirneasia org; Zambia: Food Security Research Project tomato price data, 2007/2008; USA:
h ://marketnews.usda. ov/ ortal/fv? af dm=full&dr=l& af ear id=1200002&re T e=wiz&ste 2= tru

I A: a II A I

   

e&run “ ° ‘ term? rhn‘lhm". .nme
The inverse relationship between the conditional variance results and the PPP GDP by
country was also plotted (figure 3.4). The countries with low conditional variance have
better developed horticulture markets, as proxied by the PPP GDP, have higher PPP

GDP.

73

Figure 3.4: Comparison of Conditional Variance and PPP GDP by Selected

 

 

Countries
1 o
» r.
1 E
1 8
° 8
‘c’ 2
1 .2 0
a 60
1 > a? 8
a a 'u
1: n. e
1 29 go.
1 '5 o
1 S m
1 0 a
C
1
(I
.C
1 2
3
o.
l USA Taiwan Costa Rica Sri Lanka Zambia
! finEEdfioHafi/afiaEeT—L 3113121318 FEW; pficfi Dmesfic’Product]

 

 

Source: Costa Rica: www.pima.go.cr; Taiwan: hm://amis.afa.gov.tw/v—a_sp/top-v.gs;p; Sri Lanka:
www.ggs.limeasia.org; Zambia: Food Security Research Project tomato price data, 2007/2008; USA:

b ://marketnews.usda. ov/ ortal/fv? af dm=full&dr=1& af ear id=1200002&r T =wiz&st 2=tru
e&rul‘- n D L ‘VIwPfIPEthnPCL I A' D J'L I ‘1 ILL A I

           

Of all the five countries, Zambia has the highest coefficient of variation, conditional
variance and RNPE and has the least PPP GDP. From the PPP GDP as an indicator for
economic and market development, these results show that Zambia has high tomato price
variability and low tomato price predictability which is consistent with a country that has
a poorly developed horticulture market. On the extreme end is the US which has the
highest PPP GDP and the lowest coefficient of variation, conditional variance and RNPE.
The conclusion from this is that the US has low tomato price variability and high tomato
price predictability. From its PPP GDP proxy, this is consistent with a horticulture market

which is well developed.

74

Closely following the US results is Taiwan, followed by Costa Rica and finally Sri
Lanka. From the coefficient of variation, conditional variance and the RNPE, we observe
that Taiwan has less price variability and more price predictability than Costa Rica or Sri
Lanka, but it however, has more price variability and less price predictability than does
the US tomato market. Clearly, Taiwan’s horticultural market is better developed than

that of Costa Rica or Sri Lanka.

Comparing the Costa Rica to Sri Lanka and Zambia, we see that Costa Rica has a lower
coefficient of variation, conditional variance and RNPE than these two countries. This
indicates that it does not have as much price variability and its horticulture market better
developed. This conclusion is confirmed by the higher PPP GDP (proxy indicator for

economic and market development) it has compared to Sri Lanka and Zambia.

In the earlier part of this chapter, we suggested that seasonality of supply, supply shocks,
and random variation in quantities arriving to the market are the main factors that affect
price variability and predictability. In addition to these factors, price predictability is also
affected by the absence of grades and standards. Unlike the US market which has well
specified grades and standards for tomatoes and other horticultural products, Zambia has
no formal grading system. Costa Rica and Taiwan showed less price variability than
Zambia. Each of these has more formalized grades and standards defined by either
product variety, color or quality grade. Costa Rica for instance has three different quality
grades while Taiwan has grades and standards system that incorporate variety, color and
quality. Clearly, where there are well specified grades and standards for tomatoes which

farmers are familiar with, then the pricing system in the market is more transparent

75

thereby making the farmers more confident of the price they are likely to get relative to

the overall, prevailing price level in the market.

3.5.1 Tomato Seasonality of Supply
In cases where a F FV product is faced with seasonality of production due to the
differences in the geographic production conditions in a country, markets that are well
integrated over space would reduce the severity of seasonal price variation. In the US for
instance, the climatic differences across its geographic regions implies seasonality of
production for all FFV. However, to the extent that the horticulture markets are integrated
across the different geographical regions, seasonality of supply and prices in each region

is reduced.

In Zambia, climatic differences across the country are minimal and as such seasonality of
supply of tomatoes could be the result of the size of the “market shed””. Larger market
sheds mean a market can draw from a larger area with greater variability in seasonality,
and thus reduce its own seasonality. However in the case of a smaller market shed, the
opposite is true. A smaller market shed can only draw from a smaller geographic area,
and with variability in seasonality over that small area, seasonality of supply would be

inevitable.

Soweto market in the capital city of Zambia, Lusaka, is the largest wholesale market and
can be considered as a large market shed which draws tomatoes and other F F V from a

large geographic area. Other fairly large wholesale markets in the country which however

 

17 The geographic area over which produce tends to move to a specific market.

76

draw F F V produce from smaller geographical areas include Maramba market in
Livingstone in the southern part of the country and Chisokone market in Kitwe, in the
central northern part of the country. Owing to the small size of these market sheds and
the small degree to which all the market sheds may be integrated, seasonality of supply is

a concern.

Some of the factors that influence the size of the market shed and the degree to which

they could be integrated include the following;

a. High transportation costs. Though distances across these market sheds are not
large, roads are often of poor quality, increasing the time and also the repair and
maintenance cost of transport. This coupled with high fuel costs (Zambia has the
highest petrol cost in Afiica) makes transportation of fresh produce from one
market to another very costly.

b. The lack of formal grades and standards in the markets. Where grades and
standards are either absent or not formalized, farmers or traders supplying that
market would not be confident of the prices they are likely to receive for a given
quality of their produce. On the other hand, traders who make tomato orders from
farmers would not be confident about the quality of tomatoes they would expect
from the market they are placing their orders in.

c. Poor road network. The main roads linking these market sheds are not well
maintained and as such would add to the high costs involved in transporting the

fresh produce. This is further accentuated by the absence of cold chain systems.

77

d. The absence of cold chain systems. Zambia does not have a cold chain system
which could handle the transportation of highly perishable products like tomatoes
over long distances.

e. Poor market information. Zambia does not have a market information system that
can provide general information about a product’s supply and demand situation or
the prices for the products being traded in the markets. Given such a situation and
the need for timely information on the availability of alternative markets for
perishable products such as tomatoes, random fluctuations in the quantity of
tomatoes in a market would occur very often. With market information, suppliers
could strategically channel their tomatoes to areas where they are needed and not
deprive a market or oversupply another market. In the case of tomato traders,
market information would also allow them to make their orders easily.

f. When contractual arrangements between suppliers and buyers are not met,
participants in a market would not be confident about being a supplier (or buyer)
in the market. This would particularly be the case where a market does not have a
transparent and competitive system. This problem would be accentuated by an
ineffective legal system to deal with cases of defaults. Given this, a supplier (or
buyer) would be comfortable and confident about participating only in the market

shed they operate in.

Another factor that accentuates seasonality of supply in Zambia is the fact that many

tomato growers, especially rural smallholder farmers do not have irrigation facilities to

enable them to provide adequate water for their tomato crop. Furthermore, almost none of

78

these farmers have facilities such as green houses that would enable them have better

control over their tomato production environment.

3.5.2 Tomato Supply Shocks
Supply shocks such as disease or pest outbreaks, droughts or floods also often affect the
supply of tomatoes in Zambia. This problem may be accentuated by the fact that some
tomato growers may not have the capacity to avoid or reduce the effects of such supply

shocks.

a. The use of irrigation or a controlled production environment could help in
mitigating the effects of drought or floods. Some farmers may have the most basic
irrigation technology (pump and pipes) that would only enable them irrigate a
limited size tomato field. Therefore, in the time of a drought, such farmers would
be at risk of losing their crop especially if their fields are larger than what their
irrigation technology can cater for. Controlled production environments such as
green houses are quite costly. For a small farmer to have access to such facilities,
they would have to get a loan or access credit. However, in Zambia small farmers,
who make up the majority of farmers in the country, may not have access to
sufficient credit or cash to enable them acquire such technology.

b. In the case of a pest and disease outbreak in their tomato crop, tomato growers’
access to pest/disease control chemicals may be limited due to their cash
constraints or the general limited availability of chemicals fi'om input suppliers in

their production areas. This is particularly the case with farmers that solely

79

depend on local suppliers for their agricultural inputs. In the rural areas where you
find such farmers, the input markets are not very well developed.

c. Tomato supply shocks are also affected by the farmers’ poor knowledge of how to
control pests and/or diseases that affect their tomato. This is further worsened by
the fact that they may not have access to agricultural extension services or any

early warning on tomato disease or pest outbreaks.

3.5.3 Random Fluctuations in the Quantities of Tomatoes Arriving in the
Market
Another factor that influences tomato price variability and predictability is random
fluctuations in the quantities of tomatoes arriving in the market. In Lusaka’s Soweto
market, these fluctuations have been observed to occur within the day and also within the

week.

Some of the factors that contribute to these random fluctuations in supply have to do with
the variations in the number of tomato suppliers in the market at any given point in time.
This is especially the case since the suppliers all work independently of each other and
are interested in offloading their product whenever it is ready for the market. In addition
to this, the absence of coordinated production and marketing of tomatoes among tomato
growers also contributes to this. Where farmers are more organized and coordinate their
production and supply, as would be the case with outgrower schemes, random

fluctuations in the quantities of tomatoes arriving in the market could be reduced.

80

Farmers usually adopt different marketing strategies about when to harvest their tomatoes
and about when to take them to the market. Some farmers may decide to harvest their
produce and supply their tomatoes once a week while others may decide to harvest a
similar field every other day. Considering the large number of tomato growers/suppliers,
random fluctuations in their own production (for reasons discussed above), and the
different marketing strategies they have adopted, random fluctuations in the quantities of

tomatoes arriving in the market are inevitable.

Some of the factors that could help reduce random fluctuations in the quantities of

tomatoes arriving in the market include the following;

a. Coordinated production and supply of tomatoes. If the farmers coordinated their
production and supply of tomatoes in the market, then they would be able to
regulate and manage these fluctuations. This coordination could be done through
the farmer cooperatives the farmers are affiliated to or through the formation of
marketing cooperatives which would have a mandate to plan which crops the
farmers should grow when, and facilitate group marketing of the famers’ produce.
With coordinated production and supply of tomatoes, farmers harvesting tomatoes
at a given time would then adopt marketing strategies that would make it possible
for the farmers to ensure consistent flow of tomatoes into the market at given
periods of time within the day or the week without necessarily oversupplying the

market.

81

The provision of market information on the demand and supply conditions of a market or
the availability of alternative markets. With such information, tomato farmers and

suppliers would be able to be more strategic about where they offload their tomatoes.

82

CHAPTER 4
MONTE CARLO ANALYSIS OF CONDITIONAL AND UNCONDITIONAL NET
RETURNS TO TOMATO PRODUCTION
From the analysis conducted in chapter 3, it was observed that tomato prices in Lusaka’s
Soweto market are quite variable and unpredictable. In addition to the high price
variability and low price predictability, farmers and traders selling in the market are also
faced with a special problem of unanticipated sudden sharp price declines. This high
tomato price variability, low predictability and the unanticipated sharp price declines are

a matter of concern to tomato growers who would like to make a good and predictable

return on their tomato production investment.

In view of these challenges, this chapter will seek to address the following;

1. Characterize and group surveyed farmers based on their typical yields, costs of
production, and seasonality of sales, and examine the average level and variability
of returns to the resulting farmer groups;

2. Analyze the effects of greater sales fiequency on the variability of price and
returns for each group. Tomato farmers adopt different marketing strategies and
some of them may include the frequency with which they go to the market to sell
their tomatoes.

3. Analyze the effects of producing consistently high or low quality tomatoes on the
level and variability of returns for each group. Soweto market data has shown that

better quality tomatoes fetch higher prices and are usually sold early in the day

83

upon arrival in the market. Lower quality tomatoes usually sell for less and are
sold later in the day.

4. Analyze the effects of supply chain improvements. Supply chain improvements
such as better market information, cold chain facilities, assembling and packaging
facilities, and others are expected to reduce price variability in markets. In more
developed countries where the supply chains are well developed, the instances of
price variability are not as pronounced as those that do not have well developed

supply chains.

Analysis under point 1 will establish the baseline net returns to tomato production for the
farmer groups while the points outlined in 2-4 above will establish net returns under three

different scenarios.

This chapter shall begin with an overview of the data used and data analysis methods,
followed by the results and discussion on the analysis conducted. Conclusions are

presented at the end.

Data and methods
Two sets of data are used in this chapter: the FSRP tomato wholesale and retail price and
quantity data as described in section 2.1.2 and data from a household cost of production

survey conducted earlier in the year (2008) as part of this research.

84

4.1 Household Survey
Druing April/May 2008, a tomato survey was conducted in collaboration with the FSRP.
In January through March, questionnaire design and a series of pre-tests and re-designing
of the questionnaire took place. This was then followed by the training of twelve
potential enumerators which involved reviewing of the questionnaire, role playing in data
collection and pre-testing of the questionnaire and enumeration process. Based on the
performance of the enumerators during the training exercise, ten were selected for the

actual data collection exercise which lasted three weeks.

i. The Survey Instrument
The survey instrument used in the survey is presented in appendix 4. The instrument
mainly focused on production and marketing costs of tomatoes. In addition to this, the
survey was usefirl in gathering information on the production and marketing decisions
farmers make as they try to get the highest return possible fiom their tomato production

investment. Specific data collected in the survey included;

- Farmer household demographics,

- Permanent laborers employed,

- Production and sales of crops other than tomatoes,

- Timing of planting and harvest of tomato over the past 15 months,

- Cost data on field preparation and crop management operations such as irrigation,
spraying, fertilizer and chemical applications, and others,

- Cost data on marketing activities such as sorting, loading, transport to the market,

and unloading at the market,

85

- Assets used in tomato production,
- Harvest frequency and weekly quantities of tomatoes harvested and sold, and

- Access and use of market information, and others.

ii. Survey Area and Sampling Design
Volume data from regular data collection in Soweto market (see section 2.1.2) were used
to identify the top 12 areas supplying tomatoes to the market. Volume data at the level of
each lot were aggregated to get the total volumes from each area for the period January
2007 to April 2008. The top twelve areas were then chosen and characterized in more
detail on farm size distribution (as proxied by data on individual lot sizes), seasonality of
supply and estimated volume-weighted average price over the period. Weighted average
prices were calculated as simple average daily prices, multiplied by total volumes for
each day fi'om the given area, that product summed and then divided by total volume

from that area over the period.

Among these top twelve areas, Lusaka West in Lusaka district and Manyika in Chongwe
district were chosen as the sample areas. Lusaka West was chosen because it has the
largest tomato market share in Soweto market and because its population is made up of
all types of farmers i.e. large, medium and small. Manyika on the other hand was chosen
because among the top twelve areas with predominantly small farmers, it has the closest

proximity to Lusaka city.

A total of 235 tomato growers were identified in both areas; 69 from Lusaka West and

166 from Manyika. The identification process involved the use of focus group interviews,

86

contact farmers18 (or community leaders) and/or snow ball identification techniques. In
Lusaka West, the identification process involved the use of all these methods whereas
identification in Manyika involved the use of only snow ball sampling techniques and

contact farmers.

The focus group discussion was aimed at finding out the specific tomato crop production
and marketing activities the farmers were involved in. At the end of the focus group
discussion, the farmers present were asked to list the names of the tomato growers in their
area. Where lead farmers or community leaders were identified, these farmers/leaders
provided a list of tomato growers in their areas. In the snow ball identification technique,

already identified tomato growers were used to identify others.

From the identified population of 235 tomato growers in both areas, a total of 121 were
randomly selected for the survey using a systemic sampling approach from the developed
lists. During survey implementation, however, only 102 of these 121 farmers were able to

be interviewed, 32 from Lusaka West and 70 from Manyika.

In Lusaka West, the farmers were drawn from three areas, namely Kuma plot, Star
cottage and Kacheta, while Chongwe had farmers drawn fiom five areas, namely Ncute,
Maali, Kangombe, Kapilipili and Katoba. The distribution of sampled farmers in each

area is presented in appendix 5.

4.2 Price Data

 

‘8 Lead farmers in this case were the farmers that are well known in the farming community due to their
exceptional farming abilities or the large quantities of tomatoes they produce.

87

Tomato price data used for analysis in section 2.1.2 of Chapter Two were also used for
analysis in this chapter. In Chapter Two however, average daily prices in Soweto
wholesale market were used because data fiom the other four countries was limited to
daily averages. In this chapter, we took advantage of the more detailed data set in
Zambia and used hourly average prices between 7am and 12 noon. The 6am and 1pm

prices were not used as there were very few observations during these hours.

Analysis Methods

4.3 Overview of Monte Carlo Analysis19
In addressing the three research objectives, Monte Carlo simulation analysis was used.
Monte Carlo simulation is a technique that involves using random numbers and
probability to solve problems. It ultimately results in the generation of probability
distributions on variables of interest which provide solutions to queries. In cases where
the objective is to determine how random variation, the lack of knowledge, or error
affects the sensitivity, performance, or reliability of a system that is being modeled,

Monte Carlo analysis is used for analyzing the uncertainty spread.

This technique involves the use of simulations that make use of computer models to
imitate real life or make predictions. The model has input parameters, random variables
fi'om specified distributions, and equations that use the parameters and random variables
to generate a set of output variables. It then iteratively evaluates model using new

randomly drawn values of the input variables in each iteration. By using some random

 

'9 This section draws from hmz/lwww.vertex42.com/ExcelArticles/mc/MonteCarloSimulationhtml.

88

variable inputs, rather than solely fixed parameters, a deterministic model is turned into a

stochastic model.

Monte Carlo simulation is also considered as a sampling method since the inputs are
randomly generated from probability distributions to simulate the process of sampling
from an actual population. In view of this, the distribution which is chosen for the inputs
is one which most closely matches data we already have, or best represents our current

state of knowledge regarding the variables of interest.

Once the simulation is conducted, the output generated can be represented as probability
distributions (or histograms) or converted to error bars, reliability predictions, tolerance

zones, and confidence intervals.

There are five basic steps in conducting Monte Carlo simulation. These steps can be
implemented in Excel for simple models, but for the analysis to be conducted in this

research, the Excel add-on @RISK was used. The five steps are:

Step ‘1: Create a parametric model of the form Y = f(X1, X2, ...... Xq)
Step 2: Specify a set of random inputs, Xi], X12, . ..Xiq)

Step 3: Evaluate the model and store the results as Yi.

Step 4: Repeat steps 2 and 3 for i = 1 to n.
Step 5: Analyze the results using histograms, summary statistics, confidence intervals,

etc.

89

4.4 The Monte Carlo Model

The Baseline Model

@RISK version 3.5 was used to carry out the Monte Carlo simulation analysis. For the
baseline simulation, the model used was a basic model of farmer total profit and farmer
profit per hectare. These are both outputs in the model and are functions of the inputs;
total gross revenue per trip, cost of production per hectare and the area under tomato

production.

- Total gross revenue per trip is a function of tomato prices and sales of tomatoes
made per trip.

- Tomato sales per trip are a function of total tomato production and the number of
trips the farmer made to the market.

- Total tomato production is a function of tomato yields and the area under tomato
cultivation.

- Number of trips a farmer made to the market is a function of the number of weeks
the farmer sold tomatoes in the market and the number of trips the farmer made

each week.

Based on these inputs and outputs, total profit is modeled as follows:

(6R1. )— CH * 2 (4.1)

M21

TP=

N
ll
~

90

TV __ _
TP=Z(Pi *sl. *CF)— CH*A (4.2)
.=1

51;; (4.3)
TPr = 17*] (4.4)
N = W107! (4.5)
Where,

TV 2 Total number of trips made to the market from production on the chosen field.
This is a fixed parameter.
TP = Total Profit (ZMK),

OR- = Gross revenue per trip (ZMK),

CH = Production costs per hectare of tomatoes (ZMK/ha). This is a stochastic random

variable which does not vary across trips but does vary across iterations,

Z = Area of the chosen field under tomato cultivation (ha). This is a fixed parameter.

Pi = Price per crate of tomatoes (ZMK/crate) realized during the sales trip, drawn from
the chosen distribution of prices during the season when the farmer was selling
tomatoes. This is a stochastic random variable and varies across trips and
iterations,

S = Mean sales of tomato per trip (mt). This is a stochastic random variable equal to

total production divided by number of trips; it does not vary across trips but does

vary across iterations,

91

CF = Fixed conversion factor of 37. A crate of tomatoes weighs 27kg and therefore a
metric tone of tomatoes would have 37 crates (1000kg/27kg),

T Pr = Total production. This is a stochastic random variable equal to the product of
yield and area of field; it does not vary across trips but does vary across iterations,

Y = Tomato yield (mt/ha). This is a stochastic random variable which does not vary

across trips but does vary across iterations,

W = The number of weeks the farmer sold tomatoes in the market. This is modeled as

a fixed parameter.

7’77: The number of trips a farmer made each week. This is also modeled as a fixed

parameter.

Total profit per hectare was obtained by dividing equation 4.2 by area (2 ):

TPH = TP/Z (4.6)

NW and TW are modeled as fixed parameters to simplify the simulation and because they
are expected to have substantially less influence on the level and variability of profit than
will the stochastic variables of price, yield, and cost per ha. Because these last three
variables are modeled stochastically, our output variables of interest (farmer total profit

and profit per hectare) are also stochastic variables whose distributions can be examined.

The simulation analysis of the baseline model was then followed with simulation analysis

of three different scenarios:

92

1. Selling tomatoes more frequently in Soweto market ,
2. Sales of tomatoes associated with supply chain improvements,

3. Selling high quality versus low quality tomatoes in the market

Calculating production cost per hectare initially involved a calculation of the individual
costs that go into their tomato production and marketing activities. Total costs were then
obtained by summing up all these individual costs. Total cost per hectare was then

computed by dividing total costs by the area under tomato production:

Z C
CH: 211 (4.7)
2:] A

Where,
Z = The number of production or marketing activities,
CH = Production costs per hectare of tomatoes,

:4 = Fixed area under tomato production, and

C 2 = The cost associated with each production/marketing activity.

The following were the activity costs included in this variable;

93

- Seedling costs

- Seed costs

- Field preparation costs -— ripping, ploughing, disking and ridging
- Irrigation costs

- Cost of permanent labor

- Cost of piece work labor

- Cost of fertilizer

- Cost of chemicals (herbicides, fimgicides, fungicides and bacterialcides)

Harvesting and marketing costs

Defining farmer groups - Analysis of the baseline model and the other scenarios
involved the use of four different groups of farmers. Afler extensive exploration of the
data for variables that would distinguish farmers by their performance as tomato growers,

two variables were chosen:

0 The total number of months the farmers sold tomatoes over the previous 12
months. This variable considered all tomato fields the farmer operated, not just
the specific tomato field being analyzed; and

o The season during which the farmer planted and sold their tomatoes from the
specific field chosen for analysis. Season was divided into two: the dry season, in
which farmers planted their field between April and June and sold during July to
October, and the wet season, in which farmers planted their tomatoes between

August and December and sold during November to March.

94

This classification scheme resulted in four farmer groups:

Group 1: Produced from selected field during dry season, and sold tomatoes from all
fields during six months or less

Group 2: Produced from selected field during rainy season, and sold tomatoes from all
fields during six months or less

Group 3: Produced from selected field during dry season, and sold tomatoes from all
fields during seven months or more

Group 4: Produced from selected field during rainy season, and sold tomatoes from all

fields during seven months or more

The variable, ‘number of months in which the farmers sold their tomatoes’ was divided
into those that sold their tomatoes in the market for six months or less and those that sold
them for seven months or more. T-tests for the differences in means across a range of
relevant performance variables for these two groups were computed and analyzed (table
4.1). Comparing farmers selling during seven months or more to those selling six months
or fewer, the former planted more fields, had a chosen field nearly twice the size, sold
tomato fi'om that field for 50% more weeks, achieved more than double the yield, and
had a one-third lower production cost per crate of 27 kg. Their production cost per ha
was higher, but this was due to more intensive production resulting in higher yields.
Differences in the frequency of sales and in our measure of market knowledge were not
statistically significant. Finally, farmers from the two groups were spread nearly equally

across the seasons in the timing of their planting and sales, suggesting that the observed

95

differences were due to differences in farmer resources and abilities, not to seasonal
effects on production.

Table 4.1: Results of t-test for Difference in Means

 

 

Significance
Means for Means for level

Characteristic farmers selling farmers selling difference in
<= 6 months >= 7 months means
Total number of tomato fields planted past 12 months 2.4 3.4 0.000
# of weeks harvesting from chosen field 7.2 10.9 0.000
Yield on chosen field (mt/ha) 31.6 67.1 0.000
Size of chosen field (ha) 0.28 0.48 0.000
Production cost/crate on chosen field 25,003 17,098 0.006
Production cost/ha on chosen field (ZKW) 22,352,235 33,133,132 0.007
# of sales trips per week from chosen field 1.2 1.4 0.445

Ranking on price level prediction (higher is better, max

possible=5) 1 .7 1.9 0.083

 

Other characteristics of the farmers groups were also examined based on a subset of
variables ranging from farmer demographics to specific farmer attributes concerning their

tomato production activities (table 4.2).

From the table presented, it is quite evident that there are reasonable differences in these

farmers groups.

Controlling for season, and considering the two distinct farmer groups based on the
length of time they sold tomatoes, the general conclusion is that farmers who harvested
and sold tomatoes for seven months or more produced and managed their tomatoes at a
higher capacity than those that harvested and sold tomatoes for six months or less. This
can be seen fiom the lower unit production costs they achieve and the fact that they have
higher yields and cultivate larger fields than the famers that harvested and sold their

tomatoes for six months or less.

96

Additionally, the farmers that sold tomatoes for seven months or more planted larger and
more tomato fields over the 12 month period reviewed, and during their harvest period,
they made fewer sales trips per week than the farmers that harvested for six months or
less. Further observation of other farm management practices and activities also reveals
significant differences between the two groups. For instance, a look at the proportion of
farmers that planted seedlings, it is observed that more of the farmers that harvested for
seven months or more planted seedlings (utmost 23%) than did the farmers who

harvested for six months or less (utmost 16%)

With regards to the application of lime in their tomato fields, all farmer groups had few
farmers who applied lime to their fields. However, it is observed that the farmers that
harvested their tomato crop for 7 months or more applied more lime to their tomato fields

than those that harvested their tomatoes for 6 months or less.

Most of the farmers owned that animal traction they used in their fields. Amongst those
that harvested their fields for 7 months or more, almost 60% of them owned animal
traction, while amongst those that harvested their tomato fields for 6 months or less had

at most 35% owning the animal traction they used.

A look at the use of permanent labor and piecework labor in tomato fields shows some
differences among the four farmer groups. It is noted that among farmers in group 2 and
4, who grew wet season crop and harvested for 7 months or more used twice as much

permanent labor as those in group 2 who harvested their tomatoes for 6 months or less.

97

Table 4.2: Farmer Characteristics Based on Selected Variables

 

Farmer Group

 

 

Farmer group variables 1 2 3 4
Mean number of adults (aged between 19-65 years
old) in household 3.0 3.6 3.5 3.3
Mean total size of household 7.4 8.7 10.3 10.0
Mean highest number of years formal education
across all members 9.9 9.4 10.6 10.9
Mean umber of people involved in non-farm business 0.6 0.4 0.3 0.3
Mean number of people involved in salaried jobs 0.2 0.1 0.1 0.2
Mean number of non FFV crops produced 2.0 2.1 2.4 2.8
Mean number of F FV crops other than tomato
produced 3.7 3.8 4.5 4.6
Mean number of non FFV crops sold 1.1 1.2 1.4 1.7
Mean number of FFV crops other than tomato sold 1.5 2.5 2.5 2.3
Median quantity of maize produced (kg) 3,450 2,760 2,875 4,313
Mean total area of tomato planted across all fields
(hectares) 1.65 1.65 3.34 3.21
Median expenditure per hectare on fertilizer (ZMK) 2,090,535 928,198 2,060,000 2,060,000
Median expenditure per hectare on plant protection
chemicals (ZNIK) 3,801,881 4,466,778 14,408,849 4,270,322

Median replacement costs for all production assets

owned

Weighted average percent of tomatoes that go to
waste in field

Percent farmers using hybrid seed or seedlings
Percent farmers that plant seedlings

Percent farmers using irrigation

Percent farmers that apply lime

Percent farmers that use animal traction
Percent farmers owning animal traction used

Percent farmers that use permanent labor in tomato
fields

98

19
31.6%
15.8%
97.4%

7.9%
60.5%
34.8%

36.8%

22,142,900 4,683,000

18
18.2%
6.8%
90.9%
11.4%
56.8%
32.0%

25.0%

16
35.5%
22.6%
100.0%
19.4%
67.7%
57.1%

38.7%

8,619,000 18,769,000

12
18.2%
21.2%
97.0%
18.2%
66.7%
59.1%

45.5%

Table 4.3 cont’d

Percent farmers that use piecework labor in tomato

fields 71.1% 59.1% 71.0% 75.8%
Percent farmers that use at least one safety precaution
measure when handling chemicals 97.2% 97.7% 100.0% 100.0%

 

The model used in the simulation analysis incorporated the four farmer groups based on
the variables field size; number of trips per week; total tomato sales per trip; tomato

yield; tomato production costs per crate and the price per crate20 (table 4.3).

 

20 Variable means in Table 4.1 were calculated without regard to season, while season was considered in
Table 4.3; mean values for common variables are therefore different across the tables.

99

 
 

 

 

 

 

 

 

 

 

 

£353 5888 E
88: a; «3 a case swarm?
.830 E .82 m3 5 E 082: 82am m
.8 £8 E E 3 2 N3 .wma§< mm a.
$2 .82 E 22 .883
.9; .84 as 2 .2: 9:2 .9 a m
we age... a: N0 :33 we 33% c B
........ Ea.....--..-..2._...w..aa....e.m 22::
.2383 .8 M28 cm 238 so
$8.3 E .03 a: me E 0293 .33 8am
.8 .E E 2 2 S as 03-8%. a m
22 .8: 2 A Sz ”maaE
.2 :3 in 4mm 2 .88 22 .8 8:8 be
. ...... was? new we age... .5 ago...
2383 A882
.33: won 08 43 3 5363.8 smock...
.8m 3 E .83 so as E .5 E oowwscm
m S 0 3o ”“85; 3. N
22 62 E 22 5.38
.aa .wao E 2 .88 32 .8 :3
es 332. as N8 332 .5 aao< -. ..
530m Ho 0
2333 case a
. .5 ”2 .2: £8 8o 2 w w 5.: WE
80 Q E mm :8 E E 3me .va Hm : S w moo MMMMM mm _
. 0.2 9.52 was“ . 82.8 M5
M5. : E 2 m: :a ms 52 MN
S :23 :32 .49 saga.
Q25 SE 0:28 ESE 22> 8.5 as; use 835 888 a m 3:88 E 880
288: Sn :00 v.28 mo hath 80a was—om one». Bum Bahia—SE 22233030 “men 5 0388
a 33 a? Bob 82 28 he a 9E8 2:85
38385 as . cease. Jae a .22 BE

38% ..aozv a§§> 8025 0283.288

832a> 88552 BE

 

 

Eco—Z zeta—=85 2.80 3.35 25035 ue 0.53—raw 2: .3 acreage:— emmam £6 03:.

100

Distributions

The distributions for the random variables yield and cost per hectare were identified
using the Fit Distribution facility in @Risk 5.021.The distributions for costs are
presented in appendix 6. In selecting the distributions used in the simulation analysis, a
key concern was in closely approximating the mean, median and standard deviation of
the empirical data while ensuring that the probability of getting negative random draws in

these input variables was minimized.

With this in mind, the distributions for cost per hectare and yield that @Risk ranked first
and the rank of the actual distribution used in the analysis are compared (table 4.3 and 4.4
respectively). For the cost per hectare distributions presented, in all cases, the model was
designed so that any random draw below (above) the empirically observed minimum
(maximum) cost per ha was replaced with that empirical minimum (maximum). This

procedure resulted in replacement rates of between 1% and 3% (table 4.4).

Table 4.5: Distributions for Cost/ha

 

Distribution ranked Distribution Rank of % of random draws % of random draws

 

first by @Risk used in the distribution replaced with replaced with
Farmer analysis used in the empirical empirical
_group analysis minimum maximum
1 Log Logistic Log Logistic l l - 2
2 Log Normal Inverse Weibull 3 1.3 2
3 Inverse Gauss Inverse Gauss l 3.2 2
4 Log Logistic Inverse Gauss 2 3 l

 

For the yield distributions (table 4.5), @Risk ranked first the very distributions used in

the analysis. Based on examination of the empirical yield data and on the fact that

 

2‘ Version 3.5 does not have a Fit Distribution facility. Version 5.0 was available only on campus based
departmental computers; distributions were therefore fit using version 5.0 and then incorporated into the
models based on version 3.5.

101

famers can in practice suffer a total crop loss, negative random draws for these variables
were replaced with values of zero, while the maximum was replaced with the empirical
maximum. As in the case of cost per hectare, this procedure resulted in very few

replacements (table 4.5).

Table 4.6: Distributions for Yield

 

 

Rank of % of random draws
Distribution Distribution distribution replaced with

Farmer ranked first by used in the used in the % of random draws empirical

_group @Risk analysis analysis replaced with zero maximum
1 Inverse Gauss Inverse Gauss 1 2.3 7.4
2 Log Logistic Log Logistic l 0.0 0.6
3 Exponential Exponential 1 2.5 3.2
4 Exponential Exponential l 1 .2 5 3 .2

 

Correlation of Variables in the Simulation Analysis: The random input variables in the
simulation model, yield and production cost per hectare, were highly and positively
correlated with correlation coefficients of 0.63, 0.79, 0.819 and 0.924 for groups 1, 2, 3
and 4 respectively. If the simulation analysis was carried out without taking into account
this correlation then we assume that the two are independent. When the two are treated as
independent random variables, then the result of the random draws made during the
simulation analysis would periodically result in very unlikely situations such as
extremely high yield and low costs per hectare. In reality, such a situation can not be
observed due to prevailing condition during tomato production such as poor weather

conditions and plenty pest problems.

To deal with this correlation, when setting up the model in @Risk 3.5, the correlation

coefficients for the two variables were placed in the correlation matrix before running the

102

simulation. When this was done and the simulation was run, each random draw of either

variable took account of the other and avoids unlikely situations.

Number of iterations for each simulation: Two thousand iterations were conducted for
each simulation. With this number of iterations, the confidence interval for each mean is
narrowed down. The baseline model was used as the basis of reference, and at 95% level
of significance, the confidence intervals for mean profit per hectare for the four farmer

groups were computed (table 4.6)

Table 4.7 : Confidence Intervals for the Profits per Hectare Variable in the Baseline

Model

Farmer Standard deviation for Mean profit per hectare Confidence interval at 95% level of
Group profit per hectare significance
1 16,179,220 5,043,388 5,043,388 i 709,085
2 23,447,803 19,619,038 19,619,038 : 1,027,645
3 29,107,832 25,793,930 25,793,930 1 1,275,707
4 68,247,561 50,365,066 50,365,066 : 2,991,081

4.5 Results

4.5.1 Distributions of Farmer Profits
Data for stochastic input and output variables from the 2,000 iterations of the baseline
and each of the three scenarios were copied into SPSS for analysis”. Mean, median, and

probability of negative returns were computed for each simulation.

 

22 Histograms of farmer profits per hectare are presented in appendix 7. The horizontal and vertical axes
of the histograms have all been scaled equally to facilitate comparison.

103

 

4.5.2 Simulation Results for the Different Scenarios

Baseline Results

Baseline level profits and prices for the four defined farmer groups were computed and
later analyzed (table 4.7). Results presented reveal that farmers that sold their tomatoes in
the rainy season were faced with higher costs but earned higher incomes and had much
lower probability of losing money than those that sold their tomatoes in the dry season.
For instance, an average farmer in group 2 would earn 19.6 million ZMK/ha and would
have a 16% probability of making losses. In the case of the farmers in group 1 —- the same
farmers as group 2, but selling in the dry season rather than the wet season -- an average
farmer would earn 5 million ZMK/ha and would have a 39% probability of making

losses.

Table 4.8: Baseline Results for Simulation Analysis

 

 

 

Farmer Groups
Indicator 1 2 3 4

Profit per ha (ZKW)

Mean 5,043,388 19,619,038 25,793,930 50,365,067

Std. deviation 16,179,220 23,447,803 29,107,832 68,247 ,561

Coefficient of variation 3.21 1.20 1.13 1.36

Share < 0 0.39 0.16 0.12 0.05
Average price

Mean 22,966 36,275 22,966 36,275

Std. deviation 3,645 5,173 3,084 3,736

Coefficient of variation 0.16 0.14 0.13 0.10

 

A comparison of farmers in group 3 and 4 also reveal the same general pattern. An
average farmer in group 3 earns 25.8 million ZMK/ha and would have a 12% probability
of making losses compared to an average farmer in group 4 who would earn over 50

million ZMK/ha and would have only a 5% probability of making losses.

104

Therefore, for the data collected during this period of analysis, this means that the
farmers that produced tomatoes during the rainy season when prices were high had better
returns per hectare. Also noted is that the standard deviation of profits is consistently
higher in the wet season (as expected), however the higher mean prices dominate, leading

to lower probabilities of loss despite the greater variability in returns.

Farmers in groups 3 sand 4 (those that sold tomatoes — from all their fields, not just the
chosen field -- for 7 months or more) have better returns than those in groups 1 and 2
(those selling for 6 months or less) even when they are faced with the same distribution
of prices. Farmers in group 3 earned five times more on average than those in group 1,
and similarly, farmers in group 4 earned than 2.5 times more on average than those in
group 2. The higher returns for farmer groups 3 and 4 could be attributed to the farmers’
better knowledge of tomato crop production techniques, greater access to inputs to
control pests and diseases and greater access to financial resources to pay for labor for

weeding and the procurement of other tomato production inputs.

Selling tomatoes more often reduces uncertainty regarding the average price that the
farmer will obtain. Group 3 farmers had a coefficient of variation of price nearly 20%
lower than that of group 1 and this is attributed to the 15 trips made by group 3 (facing
the same price distribution) while group 1 only made 11 trips. Therefore, on account of
the several trips the farmers in group 3 made, and assuming that the farmers are interested
primarily in the average price they receive, not the price each trip, these farmers are faced

with less price variability.

105

Scenario 1: Increased Sales Frequency (More trips made to the market)

In the analysis of the scenario of increased sales frequency, it was now assumed that all
farmers in the four different groups made 16 sales trips to the market. Instead of the 11, 8
and 15 trips made by groups 1, 2 and 3 respectively, all groups made 16 trips like group 4
did. The cost calculations used in the analysis were not adjusted to take account of the
increase in the number of trips. Total costs are expected to increase, but given that fixed
marketing costs account for a about 6% of the total costs, these have been ignored. By

doing this, the analysis is simplified and does not have major effects on the results.

Simulation analysis for increased sales frequency was conducted for the four farmer

groups and resulting average profits per hectare and price levels analyzed (table 4.8).

Table 4.9: Scenario on Increased Sales Frequency

 

 

 

Farmer Groups
Indicator 1 2 3 4

Profit per ha (ZKW)

Mean 5,179,823 19,651,172 25,856,600 50,039,507

Std. deviation 16,457,080 22,867,012 29,357,054 66,994,525

Coefficient of variation 3.18 1.16 1.14 1.34

Share < 0 0.40 0.16 0.12 0.05
Average price

Mean 22,966 36,275 22,966 36,275

Std. deviation 2,939 3,790 2,919 3,657

Coefficient of variation 0.13 0.10 0.13 0.10

 

Changes in the average profits and prices fiom the baseline were then computed and
analyzed (table 4.9). From these results, it is observed that increasing the number of trips
a farmer makes to the market does not have any effect on their profit levels since the
distributions of yield, costs and price have not changed. Therefore, the small percent

changes on the profit levels in the more trips scenario are not meaningful.

106

Table 4.10: The Effect of Increased Sales Frequency on Tomato Profits

 

Farmer Groups

 

 

Indicator 1 2 3 4
----- % change fiom baseline ----
Profit per ha (ZKW)
Mean 0.03 0.00 0.00 -0.01
Std. deviation 0.02 -0.02 0.01 -0.02
Coefficient of variation -0.01 -0.03 0.01 -0.01
Share < 0 0.03 0.00 0.00 0.00
Average price
Mean 0.00 0.00 0.00 0.00
Std. deviation -0. 19 -0.27 -0.05 -0.02
Coefiicient of variation -0.19 -0.27 -0.05 -0.02

 

The effect of a farmer making more sales trips to the market is in having more stable
average prices. It is observed that increasing the number of trips to 16 from 11 and 8 in
farmer groups 1 and 2 respectively, decreased price variability as can be seen from the
reduction in coefficients of variation by 19% for group 1 and 27% for group 2. In group 3
and 4, the number of sales trips barely increased and as a result, the price variability fell

much less than was the case in groups 1 and 2.

When a farmer makes more trips to the market, one would expect that this would lead to
more stable incomes for the farmer due to more stable average prices. However, in this
particular instance, this was not the case as very small (and not statistically meaningful)
changes in the variability of profits were observed. From this analysis, the variability of
yields and costs of production are much more important than the variability of prices in

determining the variability of profits.

This however does not mean that price variability is not important for individual farmers

facing the market, but rather it is very important because the farmers would have prior

107

knowledge of what their yields and cost of production are such that the main uncertainty
they would have to face is price. For instance, taking the case of a farmer in group 1, at
the beginning of their growing season the farmer would have some expectation on of
what their yields or costs would be, however, their expectations may differ from the
actual yields they get or costs they experience. As the farmer progresses in their tomato
cropping season, their yield and cost distributions are narrowed as they would have
started harvesting their tomatoes and would have incurred most costs associated with
producing the crop. Therefore, since when harvesting starts, a farmer now knows with
some confidence what their total tomato yields are likely to be and also knows on average
what their total costs for the crop are, the yield and cost uncertainty they are faced with
reduces. At this stage, the farmer is more uncertain about the price they would be faced
with in the market. The only control they would have in ensuring that they get a good

price for their tomatoes is in producing good quality tomatoes.

Scenario 2: Scenario on Supply Chain Improvements
Simulation analysis on supply chain improvements was conducted to assess its effect on
tomato profits (table 4.10). Supply chain improvements are expected to lead to more

stable prices that would subsequently result in greater profits.

To reflect supply chain improvements, in the analysis, the Soweto variability in prices
was replaced with those of Costa Rica’s Sa José market. San José wholesale market is
more advanced than Soweto market as it has some cold chain system in its supply chain.

Other supply chain improvements such as readily available market information,

108

formalized grades and standards, and improved transportation are expected to have the

net effect of reducing price variability in a market.

Table 4.11: Supply Chain Improvements

 

 

Farmer Groups
Indicator 1 2 3 4
Profit per ha (ZKW)
Mean 5,193,568 19,674,735 25,707,641 49,797,011
Std. deviation 16,491,907 23,271,308 28,693,658 66,772,714
Coefficient of variation 3.18 1.18 1.12 1.34
Share < 0 0.39 0.16 0.12 0.05
Average price
Mean 22,973 36,279 22,973 36,279
Std. deviation 3,144 4,556 2,657 3,412
Coefficient of variation 0.14 0.13 0.12 0.09

 

A comparison of the baseline results on tomato profits, and profits associated with supply
chain improvements was carried out. As expected, supply chain improvement does lead
to reduced price variability as can be seen from the decreasing price coefficients of
variation in each farmer group (table 4.11). However, this reduced price variability
resulting from supply chain improvements does not lead to any meaningful increases in
farmer returns since the analysis was designed to retain the same mean price but a less
variable price. The variability in yields and costs of production are more dominant in

determining the variability of profits than the variability of prices.

109

Table 4.12: The Effect of Supply Chain Improvements on Tomato Profits

 

 

Farmer Groups
Indicator 1 2 3 4
----- % change from baseline -----
Profit per ha (ZKW)
Mean 0.03 0.00 0.00 -0.01
Std. deviation 0.02 ~0.01 -0.01 -0.02
Coefficient of variation -0.01 -0.01 -0.01 -0.01
Share < 0 0.00 0.00 0.00 0.00
Average price
Mean 0.00 0.00 0.00 0.00
Std. deviation —0. 14 -0. 12 -0. 14 -0.09
Coefficient of variation -0. 14 -0. 12 -0.14 -0.09

 

Other than influencing stability in prices, some of the other aspects that a farmer would
expect from supply chain improvements include less product spoilage, better market
information, reduced transportation costs, assembling and packaging, etc. In this analysis,
only its influence on price variability was examined. In view of this, if all aspects of
supply chain improvements were considered, they would influence not just price
variability but also cost levels, and would ultimately have a greater influence on the

farmers’ returns.

Scenario 3: Quality of Tomatoes Sold in the Market

Improvements in the quality of tomatoes farmers take to the market are expected to
attract a higher price than the average quality of tomatoes would. With higher prices,
higher profits would also be expected. In this analysis, the quality of tomatoes was
defined based on their time of sale. The data used doesn’t specify high or low quality
tomatoes and furthermore making it impossible to directly compute prices for high and
low quality. Instead, the time of sale was used to reflect the quality of the tomatoes based
on the assumption that the retailers buying the tomatoes in the market would want to buy

the best quality tomatoes first. The price data further illustrates that prices on sales made

110

early in the morning are higher than those on sales made later in the morning, implying
that the better quality tomatoes would be sold at a higher price than the others that are not
of the same quality. The data also shows that over 90% of all tomato volumes arrive in
the market by 6am. Farmers therefore get into the market early but the time when
tomatoes sell is independent of the time that the tomatoes arrived in the market. The
combination of this fact with the (reasonable) assumption that retailers will buy the
highest quality first, suggests that our approach to computing high and low quality prices
is reasonable. On this basis, the prices for high quality tomatoes included prices that were
taken between 7 and 9 am while the low quality tomatoes used prices taken between 10
and 123m. Price observations between 9am and 10am were not included in this analysis

so as to ensure greater separation between the two price categories.

To assess the effect of tomato quality on tomato profits, simulation analysis for high and
low quality tomatoes for each farmer group was conducted and analyzed (table 4.12). In
the analysis, an important assumption that was made is that a farmer could get better (or
worse) quality tomatoes for the same cost of production per ha. This is true only to the
extent that better knowledge leads to better management without higher cash outlays.
However, on average over all farmers, better quality would require higher cost per
hectare and lower quality would be on average associated with lower cost per hectare.
From these results, it is observed that the quality of tomatoes has meaningful effects on

the profit levels the farmers get.

111

 

 

 

 

 

 

 

 

mod odd- dd. 8.? mod odd- cod- cod- cover—«>00 808580
5d mod 3.? vod ddd vod ad- 8d- 83333 .Bm
rod. 3 .d 2d- ddd 8d- d~d 2d- dd :32
coca om80><
and end. mmd mud- 7nd edd- mmd wdd- d v chasm
Nod mod- had vod- wdd nod- 2.6 1nd- coca-"ECO 8220300
mod. n m d w — d- S d mod- 3 d mod- mod counts—0 .3m
3 .d- 3d mmd- m _ .0 Ed- mmd vwd- find 502
$63 2 8 £88
..... 08:83 How omega .x. :1-
3:95 33 5:35 .0286 33 .3286 nwi 5:85 33 >585 .0296 33 5:95 8:865
85 85 as:
v @380 m @380 N @380 “$80
$580 683m
30?..— SaEeh. :e mum—«=0 8.38% we «8.0m 2:. "NJ. «San.
:d dd d m _ .d m _ .d m _ d 2d m d .d m fl .d ”8.538300 820580
onwm www.m Sim mm~.m 31m vow-W wand Nada 823328 .Bm
www.mm bcddm EYE $m€~ www.mm noddm Sed— memém 882
8E ow80><
mod vod 3 .0 odd "Nd m a d dmd omd d v 82m
92 Hm; 33 mo; on; N: 9;: and 5:38.»? 0:22.080
Sum. EEG H Sdgdn m3. $5.3 23.3wa m dudmdw. _ N dam-£36m w H ndmmxl d _d.~nc.: H8333 .3m
max: 3% mafimvmdm cdfidwcdd mom. Gndu dmvémwdfi dmw. d 3.3 vofimdw wnvflmdd 532
935 E .8 .85
base 33 5:88 gym 5:93 33 base base 33 £88 396 33 base 883%.:
85 85 as:
v @580 m 9:80 N m=80 _ 9:80
2380 38mm

macaw—ugh. tru—«=0 sum—m 8:. Bed me 552:5...— ”mé oEah

112

 

 

To further assess the resulting profits and price variability from the analysis of low and

high quality tomatoes, the change fi'om baseline level was analyzed (table 4.13).

In all cases where the farmers took high quality tomatoes to the market, each farmer
group registers significant percent increases in the profit levels of 37%, 22%, 15% and
16% for groups 1, 2, 3 and 4 respectively. The reverse is true for the farmers with low
quality tomatoes. The low quality tomatoes resulted in significant profit decreases in each
group, with some groups having quite high percent profit decreases such as 84% in group

1 while others had low percent profit decreases such as group 4 with 12%.

The price percent increases (decreases) are however not as large as the profit percent
increases (decreases) because profit comes only from the excess price over the cost. In
group 2, for instance, while price increased by only 10% with high quality tomatoes,
returns increased 22%. On the other hand, for those producing poor quality tomatoes,
price fell only 7% but returns dropped 14%. In group 1, this is even more pronounced
with profits dropping 84% corresponding to a price decline of only 16%. This pattern

repeats itself in the other groups as well.

It is also noted that the percent differences between high and low quality (as a percent of
the low quality price) was much less during the wet season (18.5%) than during the dry
season (25.5%). This pattern suggests that traders are less willing to pay a price premium
during the wet season than the dry season. This finding is consistent with the scarcity of

supply during the wet season, making traders willing to buy tomatoes with less regard to

113

 

quz-

Fm
his-1
wh

11]-

Fur
higl

low

Cha
prim
this

van;

The

refer

quality; when supplies are high in the dry season, traders can be more selective in what

they buy, thus driving down the price of low quality produce.

Furthermore, the impact of the quality of tomatoes on the profit levels is such that the
high quality tomatoes have registered fairly significant drops in the variability of profits23
whereas the low quality tomatoes have registered increases in the variability of profits.
The quality of tomatoes that a farmer takes to the market therefore has the effect of

stabilizing the profits a farmer would get.

Further analysis in examining the probabilities of making losses reveals that producing
higher quality tomatoes reduces the probability of a famer making losses while producing

low quality tomatoes greatly increases a farmers’ probability of making losses.

4.6 Chapter Summary and Conclusion
Chapter three highlighted problems of high variability and low predictability of tomato
prices in Soweto market. On account of the price variability experienced in this market,
this chapter examined the influence of price variability compared to yield and cost

variability on the variability of tomato returns for three different scenarios.
Three scenarios were analyzed and compared with the baseline which served as a
reference point. These three scenarios were;

- the effects of greater sales frequency on the variability of tomato returns,

- the effect of supply chain improvements

 

23 The coefficient of variation was used in analyzing the variability of profits.

114

- the effects of producing consistently higher quality tomatoes and,

The baseline results revealed that famers that produce a wet season crop get higher profits
than those that produce a dry season crop. Furthermore, the farmers that produced a wet
season crop have lower probabilities of making losses. It should be noted that these
results were obtained for the specific season that the data applied to, and there is no

guarantee that the same would happen in another year.

Production of tomatoes in the wet season is usually associated with high costs of
production due to the high prevalence of pest and disease problems and also the need for
much more fiequent weeding. What this means is that tomatoes grown in the wet season
require high maintenance. However, even though the overall costs of producing a crop in
the wet season is higher than in the dry season, the estimated cost differences per crate
between the two seasons are not very large. For instance, it cost group 1 farmers, who
were producing a dry season crop, approximately ZMK 18,500 to produce a crate of
tomatoes while it cost group 2 farmers, who grew a wet season crop, ZMK 19, 300 to

produce a crate.

Other than the differences in cost portfolios in the dry and wet season, the results of the
analysis indicate that mean tomato prices in the wet season are 60% higher than in the dry
season. Therefore, despite the high production costs the farmers may be faced with, and
assuming they get high yields, they would still be able to recover their costs and make
substantial profits. It should however be noted that there is a probability of self selection

among the farmers that produce tomatoes during the wet season: it is likely that only

115

 

better farmers would attempt to have a crop in the wet season. This therefore affects our
results, as these farmers are able to get better yields at lower cost than the less efficient or

less knowledgeable or less committed farmers who don’t attempt a wet season crop.

Concerning these baseline level results, it should also be noted that the survey was
conducted only once and this was during the period when the most recent crop the
farmers had was a wet season crop. Since recall is worse for longer periods of time, it is
possible that their recall about the (more distant in time) dry season crop was biased
upwards by their experience during the most recent season (wet). There would therefore
be value in future research gathering data on wet and dry season crops at different times
so that the recall period for each is about equal in an attempt to eliminate any possible

bias.

Examination of the results of farmers faced with the same price distributions but with
different crop management level as reflected from the length of time they sold their
tomatoes in the market, reveal that those farmers that sold for more than six months had
higher profits per hectare than those that sold for a few months. This basically confirms
the fact that one would expect a farmer that manages their tomato crop well to harvest
their tomatoes for longer periods than those that don’t manage them well. Furthermore,
the results show that the probability of getting negative returns among the farmers that
sold for fewer months was higher than the case with those that sold in the market for 7
months or more. The general conclusion drawn fiom the baseline analysis is that farmers
that produce and sell a wet season crop have higher profits per hectare and a lower

Chance of getting negative returns than those that produced and sold a dry season crop.

116

 

Analysis of the effect of increased sales frequency on tomato profits revealed that the
variability in profits is driven much more by variability in yields and cost than in the
variability in prices. Price variability becomes increasingly important relative to

variability in yields and costs as a farmer progresses through their tomato cropping cycle.

Similarly, analysis on supply chain improvements indicated that variability in yields and
cost of production are most dominant in determining the variability of profits than the

variability in prices.

The effect of producing high quality tomatoes revealed that there is a high payoff to
farmers producing higher quality tomatoes. Farmers that manage their crop better due to
the greater production knowledge they have are likely to produce a high quality crop.
With the high prices, the farmers would subsequently have higher profits and lower

probabilities of making losses than those producing low quality tomatoes.

117

CHAPTER 5
CONCLUSION

This study was conducted with the objective of understanding the structure and operation
of the tomato subsector in Lusaka, establishing the level of price variability for tomatoes
in Lusaka’s Soweto market, and assessing the impact of tomato price variability on the
returns to tomato production. An additional objective was to assess the potential role that
market information could play in improving the marketing performance of tomato

growers supplying tomatoes to Soweto market.

In addressing these objectives, both secondary and primary data were used. Secondary
data included the F SRP Urban Consumption Survey data, which was collected in four
urban centers of Zambia, namely; Lusaka, Kitwe, Mansa and Kasama, and tomato
wholesale price data from five countries namely; the USA, Costa Rica, Taiwan, Sri

Lanka and Zambia.

Data collected specifically for this study mainly constituted survey data on tomato
growers’ costs of production and data from interviews conducted with processors,
wholesalers and retailers in the tomato subsector. The tomato survey was conducted on
tomato growers from two selected farm areas of Lusaka province, namely Lusaka West
and Chongwe. In addition to the costs of production, this survey gathered information on
some of the production and marketing strategies farmers adopt, and their use of market

information.

118

Interviews were also conducted with the main actors in the modern sector of the tomato
subsector, namely processors Freshpikt and Rivonia; Freshmark wholesaler; and retailers
Spar and Melissa. These interviews were conducted with the view to gain an

understanding of their FFV procurement systems and pricing policy.
5.1 Summary of key results

5.1.1 Importance of Tomatoes
The results of this study have revealed that tomato is one of the most consumed FFV
items among the four surveyed urban consumption areas. In the four cities surveyed,
vegetables and fruits account for 15% of all food and non food purchases. Among all
FFV tomatoes are second to rape in all four cities with a budget share over all FFV of
18%. Given the significance of tomatoes in the budget share of household expenditures
and price variability which would affect both consumers and producers, further analysis

into understanding this subsector was conducted.

5.1.2 The Tomato Subsector
The structure of the tomato production and marketing system serving Lusaka is
comprised of tomato farmers categorized in three areas based on the farmer types that
dominate the area, tomato traders, tomato processors, tomato wholesalers, and a wide

range of retailers.

Ninety two percent of the tomatoes in the system come from over 150 production areas
channeled into Soweto market with a small amount into Bauleni market, while the

remaining 8% is produced by the production arm of the Freshpikt processing firm. The

119

top twelve tomato supply areas accounted for 68% of tomatoes in Soweto market during
the period January to December 2007. Three categories of supply areas were identified
namely; large, medium and small farm areas all based on the predominant lot sizes (our
proxy for farm size) of tomatoes arriving into Soweto fi'om each area. The relative shares
of each of these areas are; 35%, 33% and 24% for the large, medium and small farm
areas respectively. About three-quarters of total tomato volumes marketed in the
traditional wholesale markets of Soweto and Bauleni are directly marketed by farmers

while the remainder is sold at farm gate through traders.

The tomato system is made up of traditional (informal) and modern (formal) sectors. The
wholesale and retail systems of the tomato subsector are dominated by the traditional
sector. At the wholesale level, Soweto and Bauleni wholesale markets jointly have a
market share of 91%. At retail level, the traditional sector has a 92% share and is

dominated by open air markets and the “ka sector”.

The modern sector mainly consists of the formalized retailers and processers. The
retailers are mainly the supermarkets with Shoprite, Melissa and Spar as the main actors.
Shoprite is the largest with 17 outlets countrywide, followed by Spar with 6 outlets and
Melissa with 3 outlets. The processors are F reshpikt and Rivonia. The supermarkets and
processors jointly have a share of 9% in this sector, 1% for supermarkets and 8% for

processors.

Retailers in the modern sector all follow different tomato procurement approaches and

different tomato pricing policies too. Shoprite makes use of a centralized procurement

120

system through Freshmark which supplies tomatoes and other FF V to all its retail outlets
countrywide. It has preference for large farmers as suppliers of tomatoes as opposed to
smaller ones who are not as reliable and able to meet Shoprite’s quality and delivery
requirements. To maintain fairly stable prices during the course of the year Shoprite
offers its suppliers fairly steady tomato prices during the annual contract period they

enter with them.

Unlike Shoprite, Melissa and Spar supermarkets do no operate centralized procurement
systems for their tomatoes. Melissa has a dual procurement system that makes use of one
contracted supplier and several non-contracted ones. It has a fixed price arrangement with
its contracted supplier and this price is not altered during the contract period irrespective
of the market price for tomatoes at any given point in time. With its non contract
suppliers, the price offered for tomatoes is based on the market rate at any given point.
Through this arrangement, Melissa supermarket is able to keep its prices fairly stable

over a given period of time by averaging out the prices it pays to its two sources.

Spar supermarket is a franchise and each of its six outlets operates independently.
Downtown Spar supermarket obtains most of its tomatoes from large farmers who are not
under contract to them but are merely their regular tomato suppliers. In addition to
tomatoes fiom their regular suppliers, Spar gets some tomato fiom small farmers and
independent traders; these suppliers however have no guarantee of selling their tomatoes

to Spar when they take it there.

121

Freshpikt and Rivonia are the main F F V processors in the country with Freshpikt being
the larger of the two. In 2007, Freshpikt alone purchased 8% of tomatoes in the system.
All these tomatoes were grown on their farm. Over 60% of its canned products are
exported while the remainder is sold locally through various retail outlets. Rivonia
specializes in the production of a wide range of tomato sauces and has less than a 1%

market share

Having examined the traditional and the modern sectors, the pricing behavior of each
sector was also examined. Soweto market supplies tomatoes to almost all the retail
outlets. An analysis of the weekly prices for tomatoes in Soweto market over the period
January 2007 to July 2008 revealed that prices were quite variable. A seasonal price
pattern was observed, however, a great deal of price variation was observed within .
seasons. For example, from May 2007 to August 2007, which is the cool and dry season,
prices were as high as ZMK 903 per kg and were as low as ZMK 232 per kg. In the
months of April 2007, December 2007, March 2008 and June 2008, prices declined
sharply. From peak to trough, the declines during these periods were 60%, 50%, 71%,

and 69%, respectively, all occurring over no more than 3 weeks.

In the April 2007 price collapse, three supply areas, Masansa, Choona and Manyika,
accounted for 65% of the tomatoes in the market and were the probable cause for this
price collapse. In the case of the March 2008 price collapse, Choona and Masansa
accounted for 68% of the tomatoes on the market and are the likely cause of that year’s

price collapse.

122

Based on the price pattern observed in Soweto market a comparison of these wholesale
prices was made with four retail outlets, namely Shoprite, Spar and Melissa supermarkets
and Chilenje open air market. While Soweto market had an average price of ZMK1,179,
the retail outlets all had prices above ZMK3,000. The average price in Chilenje was
ZIvfl(3,450, ZMK3,545 in Melissa supermarket, ZMK3,408 in Spar supermarket and

ZMK3,390 in Shoprite supermarket.

Further analysis of these prices revealed that Chilenje market followed a very similar
price pattern as Soweto market with a fairly stable price markup averaging ZMK 2,284
for a kilogram of tomatoes. The pricing behavior in Shoprite supermarket followed
Chilenje market in a stepwise fashion. Among all the retail outlets Spar supermarket had
the most stable tomato price, remaining constant almost the entire period. Theessential
equality of mean prices across these retail outlets in the face of very different pricing

strategies is a notable finding of this work.

5.1.3 Tomato Price Variability and Predictability
After examining the tomato subsector and the key actors in the traditional and modern
sectors, analysis of tomato price variability in Soweto market compared to four other
wholesale markets in the world was then conducted. To determine the extent of price
variability in Soweto market, analysis of the coefficient of variation and the conditional
variance was carried out and compared with wholesale markets in the USA, Taiwan,
Costa Rica and Sri Lanka. These four countries were chosen for the analysis because of
their wide range of market development which would adequately depict different levels

of price variability in these countries. Calculation of the conditional variance was based

123

on prediction errors from a price prediction model whereas the coefficient of variation
was based on the simple measure of the standard deviation of price about the mean. To
show how difficult it is to predict tomato price collapses in these wholesale markets,
analysis of the ratio of the mean (absolute value) negative price errors to the mean

positive price errors was also conducted.

In the absence of specific information about each country’s tomato production and
marketing system, Purchasing Power Parity GDP was used as a proxy measure for the
level of economic and market development in each country. Higher PPP GDP is likely to

be correlated with the following

- Better market information,

- More formalized grades and standards,

- A more reliable cold chain,

- More integrated markets over a larger geographical area, and

- Better coordination between demand and supply for fi'esh produce

We hypothesized that countries with a well developed fresh produce market (as proxied

by PPP GDP) would experience less tomato price variability and better tomato price

predictability.

Coeflicient of Variation: The coefficient of variation is a simple unconditional measure

of price variability. It is a unit free measure of the magnitude of sample values and the

124

variation within them. A high coefficient of variation for tomato prices is an indication

of high price variability.

Among the five countries, Zambia had the highest mean coefficient of variation followed
in descending order by Sri Lanka, Costa Rica, Taiwan and the USA. This ordering of the
coefficient of variation results is identically inverse to the ordering of PPP GDP across
the countries. From the PPP GDP proxy indicator for market development, which is low
in Zambia, the results of the coefficient of variation are consistent with a fresh produce
market which is not well developed. Soweto market in Zambia lacks a cold chain, market
information system, formal grades and standards, and has small geographic market shed
for tomatoes. All these factors combined are some of the causes of the high tomato price

variability the market experiences.

Conditional Variance: The conditional variance is a measure of price predictability. A
low (high) conditional variance implies high (low) price predictability. From the PPP
GDP proxy indicator for market development, the expectation is that a country with a low
PPP GDP should have a poorly developed fresh produce market and thereby have high
conditional variance and low price predictability. Like the results on coefficient of
variation, the ordering of conditional variance results was exactly the reverse of PPP

GDP.

Ratio of the mean negative price prediction error to the mean positive price prediction
error: Unanticipated price collapses in the prices of fresh produce are characteristic of

underdeveloped markets. Due to the perishable nature of fresh products, coupled with the

125

absence of a cold chain system, there would be a tendency for tomato sellers in the
market to unexpectedly lower the prices of the tomatoes so that they sell quick enough
before they go bad. Coupled to this is the lack of market information and the poor
coordination of market supplies which would lead to periodic excess supplies of the
tomatoes. Through the analysis of the ratio of the mean negative price prediction errors to
the mean positive price prediction errors, an assessment of the unexpected price declines
was conducted for Soweto market in Zambia and the other four wholesale markets. A
high ratio indicates that a given wholesale market is more often faced with unanticipated
price declines than price rises. In such a case, operators (both sellers and buyers) in that

market have greater difficulties predicting price drops.

The study revealed that among the five wholesale markets, Soweto market, Zambia, has
the highest ratio followed by Sri Lanka, Costa Rica, Taiwan and finally the USA with the
least. These results clearly demonstrate that Zambia wholesale market is the most

problematic in terms of predicting such price drops.

Summarizing, the study found that all three quantitative indicators — price variability,
price predictability, and the problem of unanticipated price collapses — exactly followed

the ordering of our countries by the PPP GDP proxy measure of market development.

5.1.4 Baseline and Different Scenarios on Net Returns to Tomato
Production

Baseline scenario: The main findings of this analysis is that farmers that sold their

tomatoes in the wet season earned higher incomes and had much lower probability of

126

getting negative returns, despite facing higher costs of production than those that sold
their tomatoes in the dry season. At least during this year, price rises during the wet

season more than compensated for higher production costs.

The results further showed that the farmers that sold their tomatoes in the market for
seven months or more during the course of the year have better returns and lower

probabilities of getting negative returns than those that sold for six months or less.

From baseline} to other scenarios: Analysis of the different scenarios looked at the
profit distributions of tomatoes conditional on sales frequency to the market, supply chain
improvements and the quality of tomatoes sold in the market. The study revealed the

following about these scenarios;

Scenario on increased sales frequency: The results of this analysis revealed that
increasing the number of trips a farmer makes to the market does not have any effect on
their profit levels. Increased sales frequency reduces the variability of expected price but
has no recognizable impact on the variability of profits. This shows that variability in
yield and costs is much more important than variability in prices for the population of
farmers. But price variability matters very much for someone who has already raised

their crop and has a good sense of what their yield and costs are going to be.
Scenario on supply chain improvements: Supply chain improvements such as market

information, grades and standards, improved transportation and cold chain facilities are

expected to have the effect of reducing price variability in a market, and with reduced

127

price variability, the expectation is that variability in profits would also be reduced. We
proxied this effect in Zambia by modifying its distribution of prices to maintain the same
mean but lower variability, equal to that found in Costa Rica. However, this reduction in
the price variability did not lead to any meaningfirl increase in the farmers returns. The
variability is prices have very little impact in determining the variability of profits as do

the variability of yields and costs of production.

Scenario on the quality of tomatoes sold: Good quality tomatoes are expected to attract
higher prices than low quality ones and ultimately result in higher returns. The results of
this analysis have revealed that high quality tomatoes have very significant effects on the
returns of the farmers. With the baseline as the reference point, results show that farmers
selling high quality tomatoes would earn increases in profits of between 15% and 37%. In
addition, farmers also observe higher prices and lower probabilities of earning negative
returns. On the other hand, farmers selling low quality tomatoes would receive significant
declines in prices and profits. An interesting observation was that the percentage decline
(rise) in profit among those that took low (high) quality tomatoes is much greater than the
percent decrease (increase) in price, since profits come from the excess price over cost. It
should be noted that premium prices for high quality tomatoes are higher in the dry

season than they are in the wet season.

5.2 Contributions and Limitations of the Study
This research has examined the tomato subsector in Lusaka and has provided baseline
information for further work in this area. It has further made a major contribution

towards:

128

- Understanding the main actors in the system and the relative market shares they
hold, specifically, the dominance of the traditional sector at both wholesale and
retail level.

- Understanding tomato price variability in Lusaka’s Soweto market and how this
affects the tomato growers’ profits.

- Understanding of the different procurement systems adopted by some of the main

actors in the modern sector.

One of the limitations in this study was in the sample size that was used for the farmer
survey on tomato costs of production. Out of over 150 tomato supply areas, only two
areas were sampled and these areas did not include large commercial tomato growers

who definitely have difference profit portfolios from the farmers that were interviewed.

Another limitation was that it was not possible to carryout meaningful analysis on the
tomato price and quantity data to ascertain whether price information from a single
market, collected every other day, will really allow farmers to improve their marketing
performance. Analysis involving the use of price information from other alternative
markets would generate much more meaningful and useful information to farmers.
Furthermore, if the tomato cost of production survey could have also captured data on the
type of market information farmers would need and the frequency with which they would
require such information, then the analysis would be vey encompassing and provide some

guidance towards some marketing strategy that could be adopted.

129

In the scenario analysis of increased sales frequency to the market, an important
assumption that was made was that the farmers’ marketing costs remained the same as
when they made fewer trips to the market. This assumption did not have major impacts
on the analysis, since on average, fixed marketing costs were less than 6% of total
production costs; a doubling of trips for the same production level would therefore have

increased total costs by no more than 6% and typically by less than this amount.

In the case of the scenario on supply chain improvements, tomato price variability
experienced in Costa Rica’s San José wholesale market replaced tomato price variability
in Lusaka’s Soweto market. The assumption is that Costa Rica’s San José market, which
is more developed than Zarnbia’s Soweto market, was a reasonable proxy for how
Soweto market might perform if supply chain improvements were made in Zambia’s FF V
system. While this assumption is somewhat arbitrary, it is reasonable in the sense that (a)
Costa Rica has a unirnodal pattern on rainfall, much like Zambia, and (b) Cost Rica’s
system is not so far above Zambia’s present system that improvements on this scale in

Zambia would not be possible at least in the medium term.

In the analysis of the scenario on quality of tomatoes sold in the market, the assumption
that was made was that a farmer could get better quality tomatoes for the same cost of
production per hectare. This however is only true to the extent that better knowledge
leads to better management without necessarily increasing costs of production. The
analysis did not therefore take account the possibility that producing better quality
tomatoes would actually entail higher costs of production (e. g. for plant protection

chemicals) to the farmer.

130

The approach used in indentifying the two groups of farmers and then dividing them by
season of production, limited the type of analysis that could be done. Another approach
that could have been used would have firstly involved estimating regression equations for
yield and cost of production which would have included independent variables such as
level of education, farmers’ access to credit, total land size, and a farmers’ access to
extension credit. These independent variables would be included so as to establish their
influence on a farmer’s performance. The error term for each household would then be
used to identify distributions using @RISK. With the defined distributions, then
simulations analysis would be carried out to generate yield and cost of production
numbers. These numbers would have two components, a deterministic component based
on regression coefficients and values of the right hand variables of the regression; and a
random component with a distribution function fi'om the error term of the regression.
With this kind of analysis, there is flexibility on the type of farmer that can be specified.
This approach therefore would permit more interesting and flexible simulations than the

approach that was used in this study.

5.3 Future Research
The tomato survey mainly focused on two of the top twelve supply areas. Considering
that there are well over 150 areas that supply tomatoes to Soweto market, future research
could consider surveying tomato growers from the other supply areas and also use a
larger sample size. This would provide a better understanding of most of the tomato

growers supplying Soweto market and would also give better insight on how the tomato

131

price variability affects the different farmers from different geographical regions

supplying Soweto market.

This study has shown that prices are extremely variable in Soweto market and the highly
variable arrival of tomatoes into the market is one of the reasons for that. One of the
factors that could reduce this price variability is market integration. Future research could
look into the prospects of market integration in stabilizing the prices of tomatoes (or any
other FFV) in spatial markets. Such a study could focus on how cold chain systems,
improved transportation and market information could facilitate market integration in two
alternate markets (say Soweto market in Lusaka province and Chisokone market on the
Copperbelt province) when the prices for tomatoes (or any other F FV) are known. Where
market integration is possible between two alternate markets, a cardinal point of analysis
for future research would be on whether price information in the two different markets

would be more useful to farmers in deciding where and when to sell.

5.4 Policy Implications and Recommendations
From this study, some very important issues have been identified. Among them, of key
importance are the high level of tomato price variability and the dominance of the
wholesale and retail traditional sectors of the tomato subsector. Another important issue
that the study has brought out is how tomato price variability in Soweto market affects

the returns of tomato growers.

132

Tomato price variability

With regards to the tomato price variability, there are some initiatives that could be

carried out by the private sector, the public sector or the tomato producers in an effort to

reduce it. Some of them include the following;

Investment in cold chain systems. With cold chain systems in place, the
unanticipated price drops in tomato prices and the overall price variability of
tomatoes would be greatly reduced.

Local market authorities to establish formal grades and standards which the
suppliers would follow.

On the part of the tomato producers, coordination among themselves to work
towards better production and supply schedules thereby preventing large random
fluctuations in supply of tomatoes on the market. The effect of this coordinated
effort would also be in the prevention of the oversupply of tomatoes in the market
and subsequent better prices. The provision of reliable and timely price and
supply information from alternative markets would facilitate such coordination as
it would provide the tomato producers with the basis for making informed
decisions on when to produce their tomatoes, and on when and where to sell their
tomatoes.

Initiatives that would enable tomato growers access to low cost pest and disease
control inputs through collective input procurement. In line with this would be the

formation of localized tomato growers associations which would not only

133

facilitate the provision of low cost inputs but also foster information sharing that
relates to tomato production or prevailing market prices.

0 The provision of agricultural extension services specifically focused on tomato
production activities. This could be undertaken by the private sector through some
outgrower scheme which would be producing tomatoes for the wholesale markets

or a specified food processor.

Dominance of the traditional sectors of the subsector

Considering the dominance of the traditional sector at both retail and wholesale level,
infrastructure development is one of the main areas that would require improvements. In
Soweto market particularly, the wholesaling area has poor roads, lacks pavements, poor

drainage systems and has unsanitary conditions.

Through the UMDP, the Ministry of local government and Housing in Zambia embarked
on infrastructural developments in the some of the wholesale markets countrywide.
However, in Soweto market, despite investments made, there is very little improvement
seen in the market. The market still has a poor drainage system, un-tarred roads, traffic
congestion, and poor sanitation. In view of this there is still need to embark on programs

that are aimed at improving the standards in Soweto market.

Regarding the traditional retail sector, the main recommendation would be in raising the
standards of their operations and services delivery to standards that are nearly comparable
to the modern sector. This could mainly be done by upgrading the current existing

system. Some of the basic upgrading that could be done include;

134

Improved hygienic ad sanitary standards
The use of cold chain systems
Improved market infi'astructure such as pavement, roads, buildings and market

stalls where the sellers use the floor or tables.

135

APPENDICES

136

APPENDIX 1.

Checklist for Interview with FFV Procurement Managers for Supermarkets and

FFV Processors

1. Generally, how do they procure fiesh produce? Get a general appreciation for
how it is managed and the role of different types of suppliers in overall FFV.
2. Then focus specifically on tomato. Pay specific attention to these points:

a.

f.

How important is tomato in their overall FFV strategy? Is it one of the
most important fresh produce items, or are there others that are much more
important?
Do they have an internal or external procurement system? If they have an
internal one, do they do this through a distribution center?
What are the sources of supply of tomatoes? About what share of tomato
supply comes from the following types of suppliers:
i. large commercial farmers,

ii. smallholder farmer associations,

iii. independent smallholder farmers,

iv. independent wholesale traders,

v. actual market places like Soweto and others. Directly or through
agents/traders?

vi. others (describe)

What has been the recent trend in supply from smallholder farmer
associations and independent smallholders? If their share rs small, is the
company aggressively committed to increasing it? If so, why? If not, why
not?
How does tomato procurement fiom each of these suppliers work, e. g. .,
i. Do they have a list of preferred (farmer) suppliers? If they do have
a list of preferred suppliers, how does that list work, e. g., how does
one get on the list?; is it reviewed annually?; how do they decide if
you stay on or fall off?

ii. Do they use formal written contracts? If they use formal written
contracts, is it with all suppliers or only some? What do the
contracts specify?

iii. What requirements do they impose on suppliers, such as
1. periodicity of supply (weekly?) How often to they have to
make procurements from their suppliers?
2. quality standards,
3. volume requirements,
4. others.

iv. What specific dimensions of tomato quality do they require?

v. Do they buy completely ripe or slightly ripe tomatoes? Do they
buy any mix between ripe and slightly ripe tomatoes?

vi. What if any food safety practices or standards do they require?
If they procure their tomatoes directly from farmers;

137

 

i. What type of farmers do they prefer as suppliers? What are the
reasons for this preference?
ii. Do they provide any technical and financial support to the farmers?
. What is their pricing policy?
i. How do they determine prices paid to suppliers,
ii. Do they strive for some price stability throughout the year? If they
do, what kind of strategy have they adopted to ensure this?
. What geographical areas does the tomato come fiom, and about what
share of tomatoes comes from each geographical area?
i. What types of farmers operate in each geographical area.
Any other information related to their procurement of tomatoes? What are
future directions in their procurement systems?

138

 

APPENDIX 2.
Full Wholesale Tomato Price Prediction Regression Results
Chicago, United States

Table A2.1. Model Summary, Chicago Wholesale Prices

 

 

 

 

 

 

 

R 0.97
R Square 0.93
Adjusted R Square 0.93
Std. Error of the Estimate 1.59

 

Table A2.2. Table of Regression Coefficients, Chicago Wholesale Prices

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standardized
Coefficients Coefficients

B Std. Error Beta
(Constant) .182 .233
February _ 5 59* * .270 .024
March .407 .253 .020
April .342 .256 .016
May -.004 .261 .000
June .220 .259 .010
July .187 .260 .009
August .296 .259 .014
Septmnber .719*** .268 .031
October .362 .269 .016
November .743 * * * .269 .033
Decenber .082 .267 .004
Lagged Price .963*** .009 .961
Time 3.42x10'5 .000 .002

 

 

* Significant at 10% level ** significant at 5% level and *** significant at 1% level

139

Taipei, Taiwan

Table A2.3. Model Summary, Taipei Wholesale Prices

 

 

 

 

 

R 0.95
R Square 0.90
Adjusted R Square 0.90
Std. Error of the Estimate ‘ 5.00

 

 

 

Table A2.4. Table of Regression Coefficients, Taipei Wholesale Prices

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta
(Constant) 1.231 * .680
January -.329 .769 -.006
February .047 .820 .001
March -.081 .745 -.001
April -.576 .779 -.010
May -.093 .757 -.002
June 1.170 .759 .020
July 3023* * * .747 .055
August 1.781** .782 .030
September 1133* .778 .037
October 1397* * .786 .033
December -.185 .776 -.003
Lagged Price .906*** .013 .900
Time 001* * .000 .022

 

 

 

 

 

 

* Significant at 10% level ** significant at 5% level and *** significant at 1% level

140

San José, Costa Rica

Table A2.5. Model Summary, San José Wholesale Prices

 

 

 

 

 

 

 

R 0.91
R Square 0.83
Adjusted R Square 0.82
Std. Error of the Estimate 1070.44

 

Table A2.6. Table of Regression Coefficients, San José Wholesale Prices

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standardized
Coefficients Coefficients
Model
B Std. Error Beta

(Constant) 315.133’1‘III 129.635
January 413,309*** 150.452 .045
February 134.524 150.612 .014
March 26.345 148.332 .003
April -17.782 151.951 -.002
May 40.363 148.812 .004
June -l.849 149.239 .000
July 440,51()*** 147.068 .048
September -l6.398 149.666 -.002
October 321,367** 147.163 .035
November 282.215* 153.656 .029
December 513.136*** 159.495 .051
#21:?“ 353*" .015 .853
time .504*** .103 .067

 

* Significant at 10% level ** significant at 5% level and *** significant at 1% level

141

 

Colombo, Sri Lanka

A2.7. Model Summary, Colombo Wholesale Prices

 

 

 

 

 

 

 

R 0.94
R Square 0.88
Adjusted R Square 0.87
Std. Error of the Estimate 6.44

 

A2.8. Table of Regression Coefficients, Colombo Wholesale Prices

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standardized
Coefficients Coefficients

Model B Std. Error Beta
(Constant) 3243* 1.93]
January 5210*“ 1.575 -.072
February -2.576* 1.488 -.040
March -3.886*** 1.487 -.O61
April -2.613 1.620 -.035
May -2133 1.495 -.033
June -1.672 1.470 -.026
July -3.466** 1.477 -.054
August 4219*" 1.490 -.068
September 3034" 1.485 -.048
October -2.945* 1.506 -.045
December .577 1.608 .008
Legged Price .907*** .018 .905
time .003 .002 .026

 

* Significant at 10% level ** significant at 5% level and *** significant at 1% level

142

 

Lusaka Soweto, Zambia

A2.9. Model Summary, Lusaka Soweto Wholesale Prices

 

 

 

 

 

 

 

R 0.88
R Square 0.77
Adjusted R Square 0.76
Std. Error of the Estimate 276.33

 

A2.10. Table of Regression Coefficients, Lusaka Soweto Wholesale Prices

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Standardized

Unstandardized Coefficients Coefficients

Model
B Std. Error Beta

(Constant) 1 12.199 92.666
January 41.117 88.021 .021
February 1 19.905 86.057 .066
April ~58.348 86.267 -.032
May -42.367 83.220 -.024
June -169.110* 87.347 -.096
July 447.502 98.478 -.070
August -255.630* * 1 17.103 -.097
September -24.250 102.941 -.010
October -36.607 97.354 -.016
November -65.515 100.404 -.027
Decemw 444.626 1 10.876 -.058
Lagged Price .736*** .051 .736
time 1200* * * .408 .126

 

* Significant at 10% level ** significant at 5% level and *** significant at 1% level

143

 

APPENDIX 3.
Graphs of Price Prediction Residuals

Figure A3.1. Price Prediction Residuals for Daily Wholesale Tomato Prices in
Chicago, USA (January 2000 —October 2007)

 

 

 

 

0.40-
0.20-
1, 0.00- . ‘ . ‘
n . .
2 1
8
N _
u 0.20
I-
a
'u
5
m -0.40-
c
a
o
E
0.60-
-0.80—

 

1 1 1 1 1 1 I 1 1 1
01/05/00 09/25/00 06/22/01 03/20/02 12102102 08/20/03 05/07/04 02/07/05 10/21/05 07/17/06
Date

144

Figure A3.2. Price Prediction Residuals for Daily Wholesale Tomato Prices in

Mean standardized residual

Taipei, Taiwan (January 2000 —November 2007)

 

0.504

0.00-

-0.50"

-‘1 .00“

 

 

 

 

1 l l l l l I I 1
00/01/06 00/12/09 01/11/03 02/08/10 03/04/12 03/12/16 04/08/17 05/05/03 05/12/31
Date

145

l 1
06/09/16 07/06/02

Figure A3.3. Price Prediction Residuals for Daily Wholesale Tomato Prices in San
José, Costa Rica (January 2000 —October 2007)

 

0.50-

0.001

-0.50—

 

 

 

-1.00‘

Mean standardized residual

 

4.50-

 

 

 

| I j I I I I I I I I I
00/01/04 00/08/28 01/04/25 01/12/14 02/08/12 03/04/07 03711/28 04/07/26 05/03/28 05/11/16 06/07/14 07/03/09 07/10/31

Date

146

Figure A3.4. Price Prediction Residuals for Daily Wholesale Tomato Prices in
Colombo, Sri Lanka (January 2004 —October 2007)

 

1.00—

0.50—

0.00- I '

—0.50“

4.00—

Mean standardized residual

-1.50"

 

 

 

-2.00"

 

 

 

l l l l l l l 1 l I
04/01/05 04/05/21 04/09/24 05/03/11 05/07/25 05/12/05 06/04/21 06/09/06 07/01/17 07/06/06 o7/10/oe
Date

147

Figure A3.5. Price Prediction Residuals for Daily Wholesale Tomato Prices in
Lusaka Soweto, Zambia (January 2007 — July 2008)

 

1 .00‘

0.50-

0.00'

-0.50-

 

 

-1.00"

Mean standardized residual

 

-1.50‘

 

 

 

1 1 1 1 1 1 fi r f r 1
07/01/17 07/03/05 07/04/30 07/06/18 07/08/10 07/10/03 07/11/21 08/01/18 08/03/14 08/05/07 08/06/27
Date

 

—I_ yawn: 0.080.: n N .202 u _ 08: 20:00.00: :0 00m

 

 

 

 

 

 

 

oo'

 

 

 

 

 

 

_ 352:: 000: 20:00.8: :0 2:02
OER... 00 “:05 E 0.0:0000: 0500.
gal: 00...: 20: 0: 0000: :0: 8.0050001.
0:802 :008w :0 >060: .0000. 090:0
magma: 00:5: 0305 08:03 0000028: .m
F _ 005:: 6:02 600:5 5 5:83 .82 005280 e _
F r 1— _ In 00:00.2 20:00.00: .m _
_ _ .55 85:0.«1—
>003 00:90.5 ._ _

 

ago—.535 ZO~H<U~EHZHE

.omzm mmm 2.0100 0::0=m0> :20: ..5 0000:00 >00: 00> .>0>._0m $5 .0000 >0000>_:D 800m :0w_:0:2 00.:
00000000 >:0 0>0: 00> .: 00:000.: 5 000500000: OmU £08005 0:: 008:8 >0E 00> >023 2:0 000:0 00000000 0>0: 00> b 2&0: 03 >05 ”30300:: £5 E $003000“:
>: 000200 b00039 50> 0:00:00: 00> 000000.. 0: EB m_m>_0:0 0.0.0 8m0b>0 _0._0:0w 0:0 00:30 Sam—00:0 we 00:0 00=>002 000 0003 00:00:: E m0_0:0m00: 00:.0 om:
>Ew00: .00 000:0 :03 00:03: 0000.000 0: :05 0020000: 50> .005 >00 00 wcumflomram 02m >06 00> :0 650800 500000 030.00 00 00:0: >0E 00> 600300000 8 080:0
00> b .30. >0 03032—0 0:006 80.0080: 05 8 :EZmn—EZOU >AmhmquOU 000x 0: :05 00000000: 50> 09060500 :02: >._0> fl 00000000 000:, war—03.6.00 E
30: 50> .0020: 030000800 00 000E028 00:5 .80 00000:: 3:000 :3: we 0:50:00 000 00800:: 0005.: :25 0000000:_ 2 00000 E 0.00—0:00:03 00:00 000 0.5500,: >:
858%? 0:0 38.0 :80 mo wESxEE 000 5.0000000 0006500 050000 000: 0303 00:0 020:0: 0:0 20000095 900003.. 0BE0N :0 0005:0056 05 00 200000088000:
0x0E S 0000 0: 00 00:00:80: m:000:00 00 000:0 00$O 00000006 0:10:00 0:0 000.85 :000000: bt0o0w 000": 0003.0: «5&0 00000 0 .00 ran 3 >033. m0:

3:: 0. a>0>.=_m 0:00.023 020002.000 btuoom 000..— ..8 0:0..— EoEeU 000:8:—

wceu >0>u0m 003000 800.08
«00.3.:— :0:0000- >ut000w 000,:

00:80:60 000000.

«0080.505 >0E0w 3083.

.v Nun—Zak:

149

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

no
...... 0m: 30:0
m::00<
3-0:
30... 033..
00-0 own:
080:0:0
0000-:00:0m
0:0 88> 0
...... V 0050: 000
000085
>m<.:<m> >¢<.:<m7: mmHEmDmr» mmHEmsz AOOIUm ZO:H<ODQN Magz mDOKU
30:5 0:88 :3va 2:8: 8:85 0.00358 23: 80:23::
N: :00: 0:: :0>0 00:80:: N: :00: N: :00: 0:: 00>0 N: :00: 0:: 0 a: 0 0.8 .5 0050:: :05:
0:0: 00.82% 0:: :26 £5: 0003000 000800: :26 0003000 0:00:00 0800000 \0. Fm .0 0:00:0mu0b
000:: 80.: Dm:m<.:<m 00800.8: 000:: mmmEmDm >0:0b00 0:0 0 0080 00:0:0000:
0:00 000:: 0:0: 80: 8:0 1:433:02 8 >008 30:: 0080:: 00000000 :3: 8
0:0 1:<.:.O.:. >008 30:: 000:: 0:0 1:<.:.O.:. 00w0w00 EMS: : 0 :96: 00:00: 3:00:80
8 :008 30:: 8 :008 30:: >008 30m: :00: 0: 0:: 0: 3:? >008 30::
0:0:0030: 0.3: 0 20:80.: 0: 8 000.580 :0 :05 :0 8.... 50.3.0: 03:03:00.: 0:: :0 80:0 30.5.0: 30 u :0 :00 0: 00:: 3:9: 03 :.:
mU:::m<MUOEHQ .. : ZOEUHm
mo 898 28 05:53 053:0 38 0o 252 .0
00 08.8.0 200 024.2000 503%.: 0o 252 .0
we 0220800 3.0m: m2<..75|7:u 8:80:50: :0 08:2 .0
30> >52 \ 5 0000
200mm £22053: .: :
_ mPHS—MHM 38:00.00an gomnN 0.20800“: mD..:.<.:.m mmZOmwmy: .o:
”:5me: :08: 80...: 89.0.50 0:: 08000000: 8:2 :0 0802 .0

 

 

150

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

88—83
0
380mg
88:03
038i
Emma?— EE: 38: $9: .575 058 :3 89202 8 8..
”.5232 QZHU
$.me ~83.“
8059608 @6on =m B>o @6on 89A no
@388 z“ ~26 52mm cofionm 3on uofionm 58.8w So» no 88:03
no 8on m8 808% SAL so» Sana SMNV 8083 So» SQHN S’MNV vowumao 88:03 80585“

08: m . 5808 “can“ Begum han— fi=o8 SQHH Hogan 035 803 mm 5850
8 E859 «ED 580535 “:5 >828on 2 “man 05 55 vowmmqo 8:83
8:3 Bonn . 3808 >58 50>

.8ch8 88% 388 . Boa “omega no 93
Box 41? 388 ..m3 m. 9 838 so» fleeing 3.5.45 .988

8 >3 2% ow 5:8 301
wqtoflmaou v8.3-8 >5 mo 2:? 03838.59“ 08 mm “a; . 30E

 

 

D i035”?—

Azocbmm mez OH Emmy Sum
~08: >8“ “a 8.5 80% no mmmMMOB HZmZ<§mm >5 cowawuo =9» 02»: £5.88 Q “man 05 B>O flu

wohufi

E."— Hflh ZO GEOAEH mus—mogu— Eflgm—m - N ZOHHUm—m

151

 

 

Banning
8:8"2
moan—3n:
mBEoNS

Mm 8T2
BEESBHE
885:":
302"“:
2::W: :nu:
3:285 a: “sane
8:88:

:3 3 2n:
3:28:

:3 9am“:
3:88:

:3 95m“:
@0228:

:2 95:":

a: :8“:

:2 ES”:
:3 ”:2”:
:2 98m”:
:3 9:5”:
8:5 E5

5:85

 

ON
NN

ufifl Bonwmo

 

macovm

 

«>395

 

8809 803m

 

l

ooboo

 

mason .o>_o>

1

 

man 00

 

mun—G Nam—gm

 

l

83: was:

 

0838 $85

 

 

lol~ Nlmlvlm «al.: cola

0838 «a. . b>

 

28% 5.5

 

cola

GOfiOO 80m

 

 

agonaam IL
$868.80

 

uoaoqflfim m

 

BEE v

 

 

002 m

 

8: :om In

 

 

382 _

 

 

9575890

A4>m¢0

NP—ZDQMU

3980

Adm—mg

Egg

39.0 Ear.—

 

Eda—MU

>gm¢0

LOMU

 

$8an ESHN
20m #8: Sai

 

 

BMN

HEB

owe—8 805
89a gm no» Ev an;

 

“ED

 

€50

w=0m SON U6

3:93: :50: 3:3

 

onnm
mom":

398 £5
3:: no:

2: .88 :0
:a< 85m

 

E5

.535: _

€530

 

5: E:

5an :50: EB

 

oahm
mPAHH

380 m3:
Ewe/g
:9»

Ev .5893
:5? 85m

 

QED

 

 

3:3... N: 36% as :98 3353.3 3933: visas ~33: 2% 33a 39:. «an S 3.: as: 339: 8: HM

OPEOH Z<flh yum—EEO kaflU m0 man—«G :24 hmfl~5~<fl - m ZO—HUHm

152

magnum

 

153

 

 

:22. 0.35 aQDmE
9n ARE? Hui“. 30> .035 ME:

5 as»: :0: uM=h=SEL§§0 8 33:5 3
3:9: >03 .35 M3233 :33. 35:38
3% .038 05 5 33000.. M3533
a S 33.5 .3 S 0.5: 502 03 .33: 2:

..LQwQLQEREM

 

 

 

Aoanm .mo>n_v 8qu 0000: 2808 $5 $808 $5
3.88 32: as, 888%: mommommm
a: ease as: 8: E: 0220: 05:20:

:8: so: :5 mam: :58 55: 52:85
50> 30>

6:08 macaw—m

05:3: 030.5 3.2: 2%.? :03 :98 03:3: GROW 33 QMEEN was: 30>. 5:3 33: 23.50:» 2.3: 30>. «.3 S SE 20: 3:9: 8: N.»

OH<EOH m0 ENE—Lam Pmfi>¢<fl QZ< UZFEAQ - v ZOE—.UHm

154

 

 

 

 

 

this. a. QNMNK

 

.03: 30% 33:0 :000

 

3500.20: 00.3333 ..0:.::: 30..\0:. 8...: 03>.

 

 

03030:...

52.080770

.m.o£OnN H

 

 

=0 .3\ :3300 3:. :..: N: .030 3303033 .00.

 

33.8 0003.0 .23... g. .. 0. 053:0 0. 0...»...
000:. 0.00.3.3 .000.3:0 .30... 03:. 03: :0... 03.9

 

 

0.30% 02. 030:0 3.0:

 

00.30:. 0:00.: 0..:. .:..0M0 3.3.: :03:0

 

:8... six 3300 30: 0:: 30.0200

 

0:. 00: .0:. 3030:. 030:0 .N 3.03 :. .3 .:

 

0:0 30.03003 .3302

 

0.3 .3\ :..3.: M5303 0:0 0:0 .3...0:00..&

 

.3300 3.3.. 3% :30... M3033 0:0 030:0 .0

 

2.0:. .253...

 

 

:...:3 00.. :0:. 03.: 00: 30:030: 0:..3 Q

 

030303. 0.30:0 3.: 3.30:. 0:00.:

 

3:. ..:.:0.: :000 EEKQOENNkHNAflVh

 

EMNGZQQ 0:. 00: 3:. “30.30:. 030:0

 

:0:. 6:30.: M:|I...:3N 0.... 00: 30:02.0: 0:.\~ «m
“562m: ham: 5%:

53526: MIR 3.5x: .nQumSN 9*: 0:.
030:0 :0:. .2303 0:. :0:. 03.: 00.3333
:30... :.::3 0:0 3:0 00: 30:00.00: 0:.b
hQHER

N30. 030:0 :0:. ..030... 03. .3 0:0 00.33 0:0
:.:0.: «3:30 0:0 3:0 00: 30:030: 0:..3 Q

 

3.0.2.0.... 0:.3
.20.. 0:. :. ..030 0. 033K :03... 030:0 0. 30003::
0.3. 02: .030. 3:. M:.:0~E:30 .0me ”3.0.335.”

 

 

 

 

 

 

 

 

 

 

 

Bough". 2 0303080sz
28 ammo. .302 >822”.
200 .Swno Nflomum
2003.”. 005 Sean.
2.302“. 22.3”.
8.50 >352.»
200000 3.2022 2.5.59 503 :23 . .202... my:
<2 H.721:
oaum Nauv
« 82". Eu... 0.. . 8.5020... «azum
N . 3050:0025 A- 020... m2. Bonk; «sum .0
$282.00 2.2»: a. .5... no. :80 22.. as. 28”.
0.00 m3. .330: :02 Eu >HEM<> 00: 3.008 .33 00.10. 3. 22m :30...
02.3% 3 30202 050200 25.8 32. . 085E
.8 080:0 . >722 4.0.0522... 5.. 0.080. :20... 3..
Bow «3. 30m .0". .33 macaw—m 5 0.080. 0. 2.20
33 ”..0 0. .2... :02 En <15: .95»
300.503

 

5030050. 20.... a...

155

 

.333:
antsmwzu ugumxumas .3 33.85.: ixrtfifi 2:
\¢ 33 us 35 “53.8%? 2% S 235nm .8:

 

 

 

 

 

156

 

 

«N 3.3:. 3.83. aroma SouoEoch

.98.. 2: S .3 m3... :2: an .EoflOHN _
3...... Eugen. . 052282“.
2.23.3 9393 2.. ax 88 Emu... .332 .082“.
:33»... 33...: 35 835.33 v.3 08o humua aomnm
.35 a... s .3. 2.8 .2... a. 338.”. 88 22".".
.55. 256.3 0:262“. 25.9.".
has“. "2.. 8\ Q 82.... 92>an 8.50 .93....»

3.5% 3.53 .g.. 3... 33
.5433... $.38 :2... a
this.
.... hawk. .33.... .3 .... .3233
3.3.3.. 2:3 3:3: 3... 33% «N
£52... 5 .3...
.33... .3 ..3\uu~.~§ SEN Q

..LQufiLV—Ewnflm

 

 

 

 

 

 

 

 

 

 

 

.
Egg; >HHE<> $52k mam
.8308 30:3 Bo. 88 A.
$3808 9.... voomum
a as... a»... 2.3.5:.
8 have 55 Ammnfivoom .33 .3 38.33 33“... .33....
vfiqfimv 03¢... 3.2;: 3...»..3. a... 3...» =3 3...... own
3.8 mquvoomnn ..33..u........m~ 233.35....“ .3
2an voom @353 A:
38 Ban. 8.5.. but... .. 30E
~53 .o £35.08
£08 .33 so.» 2Q

 

 

UZ—HZdyw—mmZaJ—H HUM—M:— MO ZO—HdJ—s—HME GHQ Dam—m - a Swim

63»...\ 3.3.3. .3353 2.3». .3 3.... .3... unu...3...~...u...uhu.3... s... 333 33.5.»: 35.83 .3... «.3 3. 3.... 23.. 3.3.: wk. Tu.

mum—{PHD ZOEUDQOy—A - m ZO—HOHw

157

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.Qmwb 80% do
NN‘NVDSV km. 8 “can...” _
$3 2?? use: .
2a 2353 :5 fizéanm
s 32» as 8: EV anfium SE: _ 30% ”51> PE: 8% :59?» SSE mam
€83? wuguum @ Emhwufi “ED 1— 33530 BMN “ED 33530 Bow £5 no
~32: 5% as 3.39 E75 3% w< BEE
.283 50> Ev u 20E 83E
van: §E§\ui «E33» Q Hume? S80 38 06 Ev :38 Bow o 50> 38 83253
\N ..3.Eu£::fl 38 mo banana an; =u um: 9.82m
Bow cox—«E Wm
«N 93:3» .338 3%: as 8 A3» 2: “EEQNMMNNZQ ES 33%qu 289.35.:N
@6on .EofiOHS SonoEoZuc
Eowaohu: ofiugofioZum
omoo .5th SURE Noaozne _
08m Emnm aomnm
2.35an 039 who—mum
o=o>oZn> oauam<l .575 HDA<> 073an Uzqnmmdcr $575 Mwaa—
Vdo HSSMN
“moo ESL BMN Bow $5
$53 :3 8 mmwwwmmmm ¢ 22m 32m
$225 830 23> “ED Eu $588» so» 35 35055 .
. Emoo mmanwoom ofi Eu 5:8 Bow baa Bow 05 a: 9.303

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

158

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a
HmOUMEO Hmofiqflog mating—E thEUw—E ><EUAE 3042 $52,..— Hmém
45.001
mafia—m onnm
05 00080003 mo>n_ onnm
030000 50> 0?;me £038 ocuN mo>u_
E 350$ 00> tops 06 m0 wqtofimmv $000 mo>uH $5083
380 0050 >5 x003oooE m5 8 0203000 £5 3 SEN wqfisab $08M wanEE 90800 30> 5
m0 253 1?. HO H >09 00> 20 038, .803 01030020 05 20 00:5 30: 08 Sm >00 3me 3.3038
06 800.65 000on 39 “33 .00> b >5 00: 00> En— .>00 8 02 00> b 8 33 00> HQ 3.: 00> 05 0 22m 353

 

 

.02— team 3 :03: .3 on: «an

159

 

 

 

 

 

 

 

 

 

 

 

33an byte: us 8?: a»: «N.

 

 

  

 

 

 

 

 

 

 

N
N > a o m H .— N _ an? m
DEAR H25“— Ex—m Han—m Saw—m moose Nag..— anh <Hm<m <mm<h E > $52,.— mam
38 @925 Sufi
EBHN .550 ocum ouHN . H
38 RON Efihmfinm 398:0 oFmN Mbuam 25 can momufi Bum but? 2%
Es“ malfi .iu 2:? 3 >1 88L x...
H. D a. D ~22.“ .mOZ em o3 a 5 a E . u
o 05 E LSESSEN
05 "33%.:353 3:83 EA 9.“ swam .5 383 83m
9 08.: . ~38 afiguRauuB . Eu;
. a 26: man a 022.“ an“ E
. m8 . an no» b3 oqoam=SmE 0 >8 2.: Rhymxui 5 33.33 an m 50> ~
on: u on ob . a Emma mo 68 :38 so“ u 5“». an ESKuEKa 3.3 ”OMS? a vb
._ 8 uh n02: 30m Ev 5:8 Boa mo .3 so» 2Q 05 83 :33 .mozvb n 58:» 283:5.5 . H #3
.23 no» a a pa inc 05
.«o <mm< Bu 2 “SS,

 

3:39... 2% 3 338 E» 83 ES 33 ~33 as S «.93 2:3 3». r2 3.. .3 835 =: 5:: E3
.8th Q. .Eh VhV QNNNK F’s “ask. .338. «.3» E $38 3. :3 :k. :2: a. .«233 ~38 5 55:3 :23 33k. {93 .5 3:33 32 35

..MEuuauufi v.3 :3. ~8qu 2t «:15. 393%.3 “3.0.5? ..3E:= 2% 83.35 «N ..LSaxuiznm m u.

263305595: DAM—Fm - N Hm<flm

160

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmmflduv
Maximum
mawmmsoanm
mammal
mmmuaflq N
wqflmaum
mmwmwao—mnm
wimmai
www.m— mofianmm mega—5m 5:558 "Sou Fmoo moazoomm 37:5; zofiémmo :57:— mafia
“38b
BaoMnN
“38b 850$
. con ugmnm nouflono
ions oanu onnm . uofioa uofiofi .
x8380...“ maxi mural 285808 Enooom 532 B4 counoMuv $5 6:280
as 0&0on S. EOHm sq Euno
MO .500 mmaowfiono wmcouflogo mm”. 8m 68 6a noanoaO 898:2 82m
8 33 “an? 8508 20E
4.2.09 305 no 832 806 £825 :30 .9388 o 08 En
05 mm? xuoiooofi A5 no 892 288m 50> com: 20% t w v 5 . E
Ea .8» a a: a; 29 E 8.. a; 2n SEN 7: $00 is:
$3 :9» Eu 353 £335.... anonubao
cough Rana Sm Ema $5 8:980 30% Ev 303
he .5me u wanna 30> .3

 

 

 

 

 

 

161

 

 

 

 

 

 

 

 

 

 

 

”Hog—m “HUM—AH HDHAH $52."—
053 200 935 080
:05 one :0 o: 0.5 0 «.50 00 $505 “on 50> onnm
$0 a u U a H m #60 523.590 :28me 85053ku 50> mo>n_
0005x0590 50> 2:03 500 gamut: . . . .

555020 50 m0 505.8005 355 A Z .50 >to.Eoo_o 05 800 A055 30: 05005050 005095. . E55030

. . . . awash... 50 .550 «0500 0 an 05 50> H 50> .8 .555 3500 0 5” 05 50> 0Q 80:52 20E
_ . a 0 m .0 ~ _

 

 

 

 

03.3; 50505005 35.3 22.3 05 50>: 8000 gaunt: >=0_bou_m aw

 

 

 

 

 

 

 

 

 

 

 

N
HEBmOU flak: 35% DE 552,.— Ham
onHN
3.01
935 3580 as: 00 .0 EmmE so» an
Sea 5 :0» 355 >55 30: .mo> b $552535 00 Sci 853

“80 ZOHH<O§ EOE 0.5 :58 30A 9 85 20m

50% .555 3on .8 $36 a 005 50>.fi 50> 80mg 50> ED
wanna—~59: S .5: noun m:— hm

 

 

 

  

 

162

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5058sz<
fiomum
wEEwF—TN
w
suaaméfii
magi
an
«0 Huoaoomano m
383%
55%
%
.tomnm
a
w
5552558~HL
Q<HOH> m5» qu a???» @352 9:2 ><mma momdfimo wgmo ZO—Hémmo 237E Ewan
00 >5 85 5058050 8m 0 5 mMMMN
50555 my 2% ”803 5%985 25005 $00 30% “mum”: H
30H .85 5 “New_ 52: 20955 m0 «5 anmwn ”Mawm 52$ 05u~ 88255
50535 5oE>mm Mm; ._ >uw mo>n_ 55.“ 50>
m>av me u hon 5080a . 2 95 M5
9505555 85H_ o 5553555.: 5038050 % Eva SEE
38 85 852 has
o 95 u 00 .85 8382a 0:05
arms .8 0A“: a...“ 258 ”3% 505 5885 85 253 359w?” 0— >5m . 05m w.5.55
V m 50 $5.55: T5555: “Wamrmm Dumas 8» l
”5453 as >55 20 .335 so 58
5 . . no» 25
5825! “BE 8 5:52 59:. 50..“ L 05.5 3.8%

 

 

.50 9... .38 €5.08. 05 SR :05 .ESKEQJS 5&5038 =0 835.30 9......35 nu AS: 2.3.... vs
5 >30E§~ 5.2.5 8 M=§=u~fio§ut 50.509... 5.3.. 335500 450.532?qu 3va» 50.508 30.593 2% LOK§~umx-u :0 neeuinfir.» Magi-:00 §§V 2830.5..5

8mg: hwy—E OH. UZ—PEkaSfi . m mums.—

163

 

 

 

 

 

 

 

 

m
dab—Ahmoo Eta—Z Ugh Day—Hz OE": mv—HHB $52,.— H35
965 585%
585300 xooBuN 05uN
SMNV $00 > >55“ 355 5 0 # mo>n_ 5.585008 08
0505 8 5 5852 mm m 08
5 53:2 . 5.8» E. “flog » 205 323
225 a =3 58 225.0555 .25 3 SE 33 >752 305 55
055 30 50 a .5555 50> 3&5 50> _

mwwmowwmonfimomv a ”05 A”50> 55 $255 55 $5.55 205 . . 5b

 

50> 355.5 50> 35 355 >555 305 nme> b

 

 

 

 

5055 1.: 555

 

 

.55: 2.5.35

 

M55555HN.

 

mo_om

50 505802.70

 

mfiuooBHm

 

 

 

 

 

 

 

 

 

 

 

 

mm». emmnv

 

 

164

 

165

 

 

 

 

853an M £52n2 m 5.50%:00uo
ET: F 558800 "a 5 0.305800%
Emmy 22385 35% animus m ugoqaoonw 5 5.805800%
88:2 555305"? X 05505500un 975v 38:2 50552.. 5552503“
tom 0589805555 58800553; 35$ 55.855 .55< 0502": Emma «bum
30530512 8555—5 m505055m055u2 855 Z 5055545"—

uuonfitou 9:396 50. 55. 9:55.35 .82. ..0 a:

 

 

 

 

 

 

505855 2 53m .05 a A“

N

A05H~£o>u$ $8555 505055055 03585 50> 50 >55 5 KHN—A— 5.1m 5 >55 35 50> 2Q 40

3.5 .3553... 3

KDOH— Hm<mm Oh. IUDOMEH HZO Hmém I HmD A<U=ZHEU D73. KEN—Azimu— .. e ZO—HUHm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

«EZDhy—mh a . v.s—Hr.— .HmOUhu—Hh hmmhmm<z Oak—Ema...— EZEKHH— Saw—Hm Him": 232..—
. .. @625 .550
acme—um =5 b>c :0 ~55on EEMBEHN 5050505.:—
520 l €550 5n 5§=§ 55% 055: 592.5% E 51 ES 1 05530 92.2
:0» Ev 0.5052355 >55. 30: .50» 5 50m: :0»
.55 TOM—ahy—Hh ..= >420 can 55: 555 boa—5.55 n 20E
no 3%»
550» ~80 .5535.“ 25 20 5055. 30: 58: 50> 0.6 bum—5.8.5 50 8.55555 53>» 0:. =5 a: 008$

 

 

 

 

 

166

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0020350003 550 00205505 .0022 53 $020585? 00: 00000
NE SEC 5.. .0200 .2 .0... 0 5. ..0... ..0 05.50.. ..0 0.. 0... 3.000.. 0. 0.5. 0% . QMQDEED»
55.300 .000 05 5.. .305m 0.2.3 0...? .000 0... 5.000.. :0... £055.30 ..0 .020...‘ 5% tan 5.5.5 50.... 3023.50. 0....xw 0.0.5.0....Em

.50» 0000 iOEmEU 06 20 50.55 30: «5050500.... 05.80. mfiwgmwflo u 20E
0 . 50% E 005 50> 25 12.095220 5000 .20 00555555 .553 480. 5 .
EW 5: 032..
Sean... 85.0" fl.
588.5”...
can. ..
9030552 550 0502 005mu> H
Scene.”
Jgoumm 088.7 M H
ll|8§onfi .331 V.
I452? H .303" m
0.%.>u.un as.) u a
25.5" m Jag" m .20 5%.
5905" w 98852 5. 33051; 4%“
0%.": figs "an 30.
x082..um . 0385.) n. u 55.5.2030”.
8.08:”. v 8.555., u . u 55.830”...
nsomamgaozuev 8.6%.) n0 5 gamum
0200050030..nwv 205%: um u 2.540%“
0.2. 3.53 A. u .802".
03.5. . n0 pans. N 8.. gm
2; 081w 08852 0 285.2 E” 0%
28;”... 530“ N Eooaanf.

 

 

 

 

mhflmmd. 505005.. 2 n2... .05 .2 A“

A05HN.00>H 3 050.5260...“ 0.5.50.

:5» 3 658013503... 3 05.50.5555 05.50.0755 £560.55... 0.20.2550 ha 8: no. 05 .3

a: 35:20 a...

167

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

”5752—50 «ha—>50 PmOUEEU 2:05.22 Mada—5:0 h—ZDSIU RUE—*0 ”amt—2:0 552.5
50.5“.“ 50.5”.“
5550a". .352. 80.585000 .2555".
E0.wo_EHN 55.... 8.0:. =9. 0:. =0 50>0 .5 3.on 80%052um
5.0%”. 22.85550 ban. .3: 52.005550 . 80%".
5555250 .05 .0”.
55 .0 30E .5 Szo .8802“. as .0
505055055
0.05.0. .50.»

 

 

 

 

55 5005 50» .05.

 

 

168

 

BOEJ .0me

 

flapxm 85

 

SEEP

 

u=o>

 

8582 m
536 5 a: 330

 

50x0

 

30.53 x0

 

8:328 KO

flaw 5

 

 

22

 

2a a: we . 5

«Eu 830

 

 

v3 830

 

exam
ovamm>o>oam

 

 

8: cam

 

 

 

‘mo—maaflvw ' WWW?

 

 

OHEOHm Ema madam: BHZHmOU HmD< $572
an .. a 53.5.:
mi 58:» 5&3
om: 8:0 mg 725 .53....53
wmmq menace 0388.3 .5»: Re 3.3
8: 3:0 aim 3E 53: u . onnm 2: 5 :3 .55
5.? r8.» :5.
Som< 83:80 oEEoHnm CSMNV $83 8% mo vac 06 us 83%;. 3.3: man“ 3.2% 2&5?
om: 3:0 mini EH £6 58558 and 9 BNZ mm x 22a§£==m .9.»qu 2: 3:3
mMOE 3:580 Smfiokum 53.2 o.“ a?» can 05 82b maouosvoa 3.qu mafia
82:9 no M520 : ow: TH bye 5% :oz $.39» 9? V d2 _ :8 0389 v.33»? 3.. 3K
~855on o oflfi Bo E Hommm £5 Wm: M“ Show 50> Esau—aztm
0382 3 com: : mm 25“ w a m mo cm: 58: . E Emma m5 ~96: no»
Emma E6 >3 8
05 mo 5:8 Boa Eons do 05 mm “2;? . cm: :9» 0Q ow 58% 806

Emma £8 83 new use 398
ca 4;? 30% 335:9

 

 

 

“moo : 2:03 “an?

mo baa Box

 

 

 

 

 

flOfi—Uflfich Canaan LOH 02— 5° 30mm“ On: a

“03¢.

 

 

=2? .5 via 8 :33 >5: 03 .F

ZO—HUDQOw—m OH<EOH Zn DHmD mhflmm< - h ZOE—Dam

169

 

88.8 won 083E

 

gag—o “Cm— ”0803

 

80:: mum

 

=03

 

 

9:0 0? — How hone—M

 

 

a .........
m 09C. I 38.“— U>m

 

 

a ....... v
N 0&3. I won:— U>m

 

 

monofi .......
_ ugh. mega U>m

 

 

a ......... V
m 23%| 88¢ 30m

 

 

a ......... v
N 0&3. I 389 bom

 

 
 

wognva ....... V
_ ugh- mom—n bom

 

 

a ........ V
m 093. I 38% 8:888?~

 

 

AWO‘OE ......... V
N on»... I momma 888852

 

AWOHMOE ....... v
_ 09¢. - «698 8388:?

 

:88
8 “389 8888 0888

 

m8=m 983;

 

 

Amognva ......... v
m on: I 388 20885

@0808... ..V
N 093.1 888 ofionouom

 

 

H Rafi—W 88a 2.0885

 

menu o—OJOHOQ

 

Bonaoml

 

 

85 .885,

 

 

 

 

 

 

 

 

8:88

 

170

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 v m 0 00>?
N: v a 0 00>?
8 v 0 00 3
.8 use dImm 3.80 33...
08420 30 0.538 080 030 0.80 0330033 303.830 038.533 0582 0980 :57;
WWW” oanm 8308nm 0:“ 8308nm
080: 08 00083080 m0>u_ x003u~ 00080 n HESOmAL 0.003um 8080 u 8:08nm
885 :8 Bow >001 menu— >80n_ xooaum
30533.5 0280 $8.83 5003305 >0003c00m >00H_ 008: m 0 a
mo 8 8303. 88 E880 . 20800308 $053 33033003 “00:08 3
00 30> 0m .0 08
0888 08.38 weeks... 3%08 8 so» as 8 .8 05.30 583.8 83 :80 25 u 2 .0
0 0802 8 m0 0 00 30> 5 30> 3 00080 >88 32;
m . E = 0. 0.0
008.88 08 m0 080: 00089 80>
08 030 00083020 15.“ 0808 0.0—0m 30:08.05 0008009 080 88030
28 30> 8038 308 30> 8Q MOE 30> Ev 5&0 30m
8868 00008 .003:
20.33 88E

€03,028 80030080 08 380 8808 «0 000.833 338 05 00.“ 890—00 may .80 Am 6302 30.5 9» 0083 AN 8302 830.5 av
0083 00m 003: 08 80 :8 308 .2 8302 800.5 3.0 008m 00» on: 08 80 2m H02 00 .2808 8w 08 08 800 >285 v21? 00083 003 0088 05 030 2a 308

.0000 08 8 88 b .0083“ 38808 >80 :82 00800308 :5 .8308 98m 08 00m 0003mm 3:3 080 30.508008 >200? :82 888 800303000 08 68888 003
.8808 303M 0 .50 0.08— >E808 030 >383 808 8 =m 3.02 23080 30> H30 .8080 00m 38308 080 8808 080m 00.“ 000300000 >383 088 2.4.0
30 > .8 :88 0000 >08 005308 .8000 >8808 00 >200? 0 :0 0003mm 00:3. 030 30808008 .805 08» 0. 800303.000 08 030:0 3200 038 03:. £83083§~

M88338 :03 038‘ 0.: 330% 0.5 05 80.8300 333 3020.23& 303.85 8338.? 00.3.5830 05 333 :0... S 55 S .33 i0: 330.: 0x: NN

harry—{m 1707:”— Oh. HmEfl<l HEM—m I v Hm<=m

 

 

 

8508 x0 .85 2300

 

 

 

 

 

 

 

 

003mm.—

  

waa—DEO— E—

m 38826

 

 

171

 

 

m 582 3.80

 

00003

 

m 0 00m

 

Mu «DOB

 

V3

0003

 

if

 

0 8302 880.8.

 

_._
r \l

8.0003

 

H0003

 

 

.8 0003

 

u333

 

\:\
Villl‘r
V'V'VrVr

-A

 

m 808082 880.8.

 

3 08003

 

'\.v..
8.8.8.8

\|\I

n 08003

 

a 08003

 

 

\‘
.A
\I
V'VVV'

 

VE3

 

 

N 8.3082 850.8.

 

\‘
Vth

v 883

 

.. 08003

 

a 08003

 

 

V'VrV'Vr

 

 

883

 

 

 

8 £82 8808

 

 

 

 

\
' V \‘ \‘ \‘

V

v 8003

 

 

Hm—
mac—>80

EOE
MU

OHH30w

038
.330

181—Hm
.8.me

181—«V
Adm—w

Ofihnomm navy—.8830

MOT—Ogre

M08582

Ham

2378'.—

 

£08338
088 .80
080: 080
08.83 20.
0.80.8385

080830080
80800
00880
.80 0 880.8.
8008338
.80 0502

 

808333 088.80 0803

088 8033 680830080
80800 30> 8038 308
00008038 00008.8 .803 88

03nN
00>n 8

85:03
2.0 0550
OHmBOm

38 0083

03.3808 30>
88¢. 00808
30> 88D

 

finofinm
08003HN
>080” 8
530380088

00880 n

 

 

$3 3% 20
0080.80 >538 308 .803 .88

088nm
80

Hm3OmAIV
00>” 8

8.800800%
30> 8088 880

 

8800 30> 805

fiaofinm
080.80 a

>08038008m

 

 

8.8.0.588
as 0558. $052:
30> 8880 008080 >305 308.8

finofinm

08003HN

3.0L
530380038

 

383° 0

 

 

83m

 

808038005

 

3630.8 2.3 038%
068.8 8.8 20 88° 308

 

82:

803.88 .08
808 038.8.

 

 

t. 88088

 

 

172

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

88qu 8qu Jon :

 

 

 

 

 

 

 

 

 

 

 

 

 

 

we w 582 38h
v 4v v u 33
H v m u 83
n v n u 83
Av u 00>?
r V n 582 ~83.
E». v :33
Kw n v 33
Ev n u no?
2. u on?
iv o 6:02 38,—.
Av v v Av v co?
n v v H u 83
a w v n v 83
v v u on?
:— EOH 0mm 5mm a1
go do OBEOm Adm—m Emu dam—m OMHDOfih QOKQHMU MUEOE MOT—z H39: 5573
WWW E ammm 04% if
ESE 06 80:38? menu— : $va 835 u hWBOmAL x003n~ 8:20 u finofinm
MHWUMHNV awe—MW $.0ch >826on mor_ 35%“an xwmew mead mo u
w. . . on he 262 825 H
We u :89 25 5:6 Ea .5» EV bumaog m 08: a m 32$ 3 a n 20;
90x38 GEMS/Om moss baa Boa Jon: :0 35 EW m3“ 5:6 Ema: Sow 08E.
mo 0:82 5 £43 :8 no.» ED :0.» Ev 38.8 >58 Bo:
gov—HE 06 mo 082 0368 Box
ufi was 30:33? 4?. 838 mofim nouns—USA finer—on m5 macaw
28 so» none 30: :oz 20 MOE 50> Ev some 303

 

 

173

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N3. N ”4003
3v _ x83
3 m 5:02 :38.
vmv w x83
mmv m x33
va N x83
Ev _ x33
Nv N 5:02 50H
«NV w x8?
va m x83
NNv N x83
_Nv _ #003
2. > 5:22 30%
iv v x83
m S. m #33
N? N :33
0 :v fl x83
. .g.—M EH08 O my.— Admm NUS
:0 m0 EMEVOm 15mm PKG Ada—mm Gan—Om.— GOy—hhy—U Omn— v—U—mz Hmém 557E
00x38
05 mo
08% oanN
05 Sum 585% 65383
203383 mo>n~ ”.39an fiaofinm
895 8:20 u mural 8220 n
2898 30m . >mun~ ficofium x33nN 88:
252.3 8:20 $5ng >oao=co£ 23305.8: xooBHN >€u~ no #1
mo a BBB m3“ mafia =o> 35 >31 >o=ozco£ “meta: ~
Chg/Om 82m 3 u 20E
8338 am =a =3. 30> ED >826on 89a 08;
go ”we 2% 8% 29 835 255m 25 mane Emmi/E
0:82 50> >58 Bo: Jon .3 30> Ev W380 >58 30m
3x88
08 mo can: 05 28 JAM 8:2: Quorum
6585020 28 30> no 20 83m 2052605 mEu mug—v MOE
n88 Bo: 38me 30> Ev undo 30m

833 .8: b

 

 

 

 

 

 

 

 

 

30mm vacuum

174

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5v R. 5:02 50:;
vnv v 0:003
whv m 0:003
Nnv N 0:003
:v : 0:00?
0v 0 5:02 30:;
vov v #003
mov m 0:003
Nov N 0:003
Ev H #003
mv m 5:02 530%
vm v v #003
mmv m 0:003
va N 0:003
:mv : 0:003
vv v 5:02 30,—.
vvv v 0:003
O mvv m 0.003
Humv— EBOH O m:— ‘Emw MU;
$29 m0 REBOm Adam BED Adda—Hm Qumran—Om.— GOQA—HKU Ow: M037: Hg.— 2375
90010::
05 :0
0:3: 0:“
20 2E 388nm 6538 AM
8055050 00>": x003HN 5:0::Hm
8:3 83:0 03 00>": 080:0 u
20.:20 200 . >05": 5:0Enm x003nN 00:5
352.3 00:05 052:0: 3:03:08”: 2030020: x003nN >5": :0 t
:0 u 530:. 05 $3.33 30 > :05 >35"— >0:03v0:m 5030: _
8010:: Ohms/Om :0 :00 30> ED >0:03¢0:m 00:33 80.5 080?. u 203
:o MMM Ea 8: 20 8:8 35:2: 25 0:96 5:33:
0:8 Z 30> >82: 30:: .8: ,3 30> 55 003:0 >:0:: 30m
:0 :08
05 :0 ohm: 05 5:0 13“ 008:: 050:0:
6:05.550 20: 30> 30 BO .855 55050:: 55 w:::3_0 MOE
503:: 30: 0:355 30> 55 :050 303

0000—: .3: .:

 

 

 

 

 

 

 

 

 

175

 

 

  

 

 

 

 

 

 

 

 

 

N :03 :02: 0: :00m 0: 035 Aqmm OH 0:00:03: 5:08 :00: 0: 03m: GAE“:
03:: HZDOUwE DE 0 mqm<ZD 30> 0:03 05 0: x00: 05 mg 580020: 003 a :0:..:0 WE Z: m><m5 3030::
003000: 3010:: 05 0: 30:30:: 05 0: 0:00: 30> :33 :0 0800 090003 05 :0: :0 20m: 05 5 :05:0 30> E: 0005: u 20:”:
0:00: 30> :05 E308: 30> :05 33080 05 :00 0: 050:: 83:00:: :30> 9:300: :0: 0:800: 8:03:08: :30> :: 80:: 08F
05 :0 £1003 :3on< :0 .x. :23 :305: h00> .2 :0>0 30> 0:03 03H 30:. 05 :0: 00005 :0 Q1003 505::
5:03 80:20:: £030 M3230

 

«3.5030 .32: 0: 0:3 :.: 50:. 0: 033:: 0:02 :06 0:00.: 0: 30.: 8028.? 533:0 :05: 38350853: 30:»: 0: 0:3 3:9: 0.: 0000:0000 :0:.:.30:\ 05 3% mg.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

wv M: 5:02 :Soh
v v v 0:003
n v 0:003
N v u 0:003
:v 0:003
REM EH08 O OE JAM; MOE
20 MD hHBOm Adam HMO 41701—155 OMhQOw—m GOMEHKD Om.m MUEZ mg.— 2373
Gov—:08
05 .:0
0W0: onuN 5:0:Fn A 0:qu
5 20:30:? 00>”: 0:003HN OHWBOmAI 5:0::um
03:3 00:80 :0 00>”: 020:0 u
. 20: >00": 0:003HN 00:5
L330 5:0Enm .
00:0:0 200:0: 5:03:05 >05”: :0 :0
5.3.3 o 0 m. 0003020: 0:003” 3:03:00:
: 0 :0: H 0:5 :50 30> :05 >00”: :2 $030.: :
Ohmk/Om 000:3 a 0: 20:":
60:30:: E 00 00 :0 :00 30> 05 3:03:05 ::0:.: 08E. .
:0 0. 08.0 E00 30 E: 00:0:0 002:0 0:5 5:3: Hmm>m<m
0::02 w30>: >:0:: 30:: 00:: 30> :05 00:0:0 >58 30::
00:08 13:: 8:2.
05 .«0 080: 05 0:0 30> : 000:0:
6:230:30 20: 30> vh— 00_0w 5:03:95 05 mg: MOE
003:: 30:: 0:00:05 30> :5 03:0 30::
0000:: .:0: m:

 

 

 

 

176

 

”22m BEE
A323 3QO E entomov 883 ”835835 SEOHm
08: 088 05 “a Son 805 28 55 .quiomm 95 880:8 DEBS TN
:8 o» mmafloi 95 no.“ :53 56% H ”mafia.“ @020 “can 80383 05 28 a??? 3an H1

D $0323 we?“ 59» 39:83 “won 30anme £033 £830va “a: o? way—om 05 8h v.5

 

w 58:

co
VT

 

h 582

l\
V

 

o 5521

\O
V

 

m £82.

V3
‘1'

 

3“

.V 582M

 

- 582

 

\‘Il

.m £821

 

as:

 

 

w 582

 

F 582.

 

o 582

 

w 582

 

inch“,

AV 582

 

H as:

 

\V\'

.W 882

 

VVVVVVVVVVVV

 

 

 

W 682

 

NZ

«.5.— SB 53oz 84mm 283:. EU; $33 $575

 

177

 

 

 

@Bomm .550"
bmgv “SEE hoomuv swung—.53 55
2:88 08: no xoi 9 ofimabnm 50> 8 326 “Saw
:6: AL =u 35305 “mom Ho 083% 3.38 so» Mo 98823
28 £333 .oanm 9 $8.555 Janene gonna Sam 8%?
53:“; menu“ 58 o>mmmooxoug 8 :53 5:55

 

 

 

 

 

 

 

 

 

 

 

38 550 NH

 

 

323E800 83m :

 

 

362888
33m 98 wagons

O

 

 

A—
v

mangoes

 

 

 

mauommsfi 2a wage;

 

 

980 65% fiommadb :30

 

 

commas 95m

 

 

%;\Ol\""

mag?

 

 

_;
n v

 

 

 

wfitom .u

 

 

mini

 

 

 

HEmOn. DZHmOU
ihOHmOU mEZDHU .5750 HmOU

 

gnaw .550 u~ u

25 “on US . 08.3 380 “on no
32 mufim Hon um .88 En Mn
65: “on Mo Jan? Hon Hm

 

 

 

 

 

 

GEE—5L8”: E. E nw . v.83 “a um
um: 3m MN .383.“ Sn NH
38 Eek magic 595:: Bow “a: 92E moo
aw _ on 30 600 “moo mo 0&9 oven. “moo

 

 

 

 

«35?: a: «:53: 3:

233.3%»; 2% .33.: «3:3 3:33 35 83.35 £05 EEZE ”2.53 933:3: 3.5: 33.35 35.35 .3 .63.»
3.9.35 2% 333 .33. «.3 S 3.: 3:9: 9: Eh 33333 333.9558. 33 M533 5...: 33.33%: .339 39.35 as :33 33A. .32 SS 8 3.: :3: 3:9: 3:
g 23h I 380 ware—.32 can Maugham ad

 

 

 

 

”22m vacuum

 

 

 

178

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

n
v
m
880 .550 2
“813.3550 83m fl H
nommmunaoo 33m 98 53015 S
mgmofib m
Etc mush Ba 3934 m
tommqah :30 N.
to «SE 32:] w
3381 m
9&8 atom Eur v
flown m
mafia a
magi
Hgfimoo DZPmOU
jHOHmOU $57550 .5750 HwOU
ago u:
3.5 you HS . 08$ 880 Ba no
32 813 you "w .28 Ba MN.
£8: Hon "c £83 hon "n
Ame—7559308 E. E nv . x83 2 um
on: you MN dew—om Hon NH
38 30,—. 3.2.33 58.55 Eek “ED AMENV “moo
ago 58 mo 09C. 2.8 $00

 

 

 

Eu. Q2995. 2t .3833: 3.: 3333 3.. 35 .33: 5.53 25:32 mi 83.35 ”5: Vflmsszw
M333 M53313: 3.3.: 33.35 23 us .Mamumnufiu .33 3.8.x 3: 333353 2% .3 a3.» 38% 35 333 ~33. «a: S SE 339: 9: 3:2
9. 22m . 380 «53.52 E: «.585: 3

 

m
w
m

 

 

 

 

 

 

 

 

 

 

 

179

APPENDIX 5.
Distribution of Sampled Farmers

Table A5.1. Distribution of Sampled Farmers

 

 

Number of farmers Number of farmers Number of farmers actually
Area identified sampled interviewed
Maali 31 16 8
Kangombe 2 1 9 8
Nchute 12 7 7
Kapilipili 74 35 35
Katoba 28 13 12
Kacheta 43 28 25
Kuma plot 1 l 6 5
Star cottage 15 7 2
Total 235 121 102

 

 

180

 

APPENDIX 6.

Baseline Distributions for the Random Variables Cost per Hectare and Yield

Figure A6.1. Distributions for Cost/Ha, Group 1

-8

P’

o
A

Values x 10"

Fit Comparison for Dataset #15
RiskInvGauss(30356963,109181684,Risk5hift(-7115590))
4.1

59.9

 

.N
m
r

 

.N
o
.

 

 

 

Values In Millions

181

I

Minimum 175550000

Maximum 6582449000
Mean 2324137339
Std Dev 15188056.61
Values 38

— lnvGauss

Minimum -7115590.00
Maximum + '0

Mean 2324137300
Std Dev 16007097.40
Values 38

Figure A6.2. Distributions for Cost/Ha, Group 2

Fit Comparison for Dataset #16
RiskLogLogiStic(115281,17661041,2.5951)
53.7

‘-

Minimum 3417901 .00

Maximum 7698163300
Mean 2207992593
Std Dev 15577011 .23
Values 44

: _. LogLogislic

Values x 100-8

Minimum 115281.00
Maximum +'°
Mean 2296171566
Std Dev 2495582300

 

 

 

D O O O O O
N v ° 8 °

('1 in
Values in Millions

182

Figure A6.3. Distributions for Cost/Ha, Group 3

Fit Comparison for Dataset #17
RiskExpon(30038668,RiskShifl(3341010))

 

 

5.3 115.8
90 0%
91.4%
3.5-
3-0 ‘ - Input
Minimum 431000000
2-5 Maximum 12396190500
Mean 3434866819
:9 2.0 Std Dev 3101188556
3 Values 31
X
g 1.5 ‘ _ Expon
>
Minimum 334101000
1.0 ‘ Maximum +-°
Mean 3337967800
Std Dev 3003866800
0.5 i
0.0

 

0
20
40

a 8 8 a
Values in Millions

140

183

Figure A6.4. Distributions for Cost/Ha, Group 4

Values x IDA-8

2.5 -

N
o

E"
u:
.

E"
o
.

0.5 -

0.0

Fit Comparison for Dataset #18
RiskExpon(29852516,Risk5hift(2002645))

115.7

 

 

 

° 8 3 S 8

Values in Millions

100
120
140

184

- Input

Minimum 290726700

Maximum 13558839500
Mean 3275978348
Std Dev 3255465351
Values

_ Expon
Minimum 200264500
Maximum +-°

Mean 3185516100
Std Dev 2985251600

 

Figure A6.7. Distributions for Yield, Group 3

Values x 10"-2

1nvGauss(82.248, 162.237) Shift=-11.293

1.43437 g , 7. ,,

1
1 ‘ 1
1.2—1 1

    

 

 

185

 

APPENDIX 7.
Histograms of Farmer Profits per Hectare under Different Scenarios

Figure A7.1. Baseline Scenario: Histograms of Farmer Profits per Hectare, Group 1

MTG—7 7’ 7' ”7 7" ‘7 1

Count

 

0.00

5000000000

Profit/Ha, Grow 1 (ZMK)

10000000000

186

 

Figure A7 .2. Baseline Scenario: Histograms of Farmer Profits per Hectare, Group 2

Count

   

0.00

5000000000

Profit/ha, Grow 2 (ZMK)

187

Figure A7.3. Baseline Scenario: Histograms of Farmer Profits per Hectare, Group 3

zoo-T—

Count

 

 

0.00

5000000000 10000000000

Profit/Ha, Grow 3 (ZMK)

188

 

Figure A7.4. Baseline Scenario: Histograms of Farmer Profits per Hectare, Group 4

Count

 

0
6000000000

0.00

50000000. 00 100000000. 00

Profit/Ha, Group 4 (ZM K)

189

 

Figure A7 .5. Increased Sales Frequency Scenario: Histograms of Farmer Profits per
Hectare, Group 1

 

   

 

 

45000000000 0.00 5000000000 10000000000

Profit/Ha, Grow 1 (ZMK)

190

 

Figure A7 .6. Increased Sales Frequency Scenario: Histograms of Farmer Profits per
Hectare, Group 2

200

 

150‘

 

 

 

 

0‘1

5000000000

0.00

5000000000 10000000000

Profit/Ha, Grow 2 (ZM K)

191

 

Figure A7.7. Increased Sales Frequency Scenario: Histograms of Farmer Profits per
Hectare, Group 3

200

 

150"l

 

 

 

0.00

5000000000 10000000000

Profit/Ha, Grow 3 (MK)

192

 

Figure A7.8. Increased Sales Frequency Scenario: Histograms of Farmer Profits per
Hectare, Group 4

 

Count
s
l

 

 

5000000000 0.00 5000000000 10000000000

Profit/Ha, Grow 4 (MK)

193

 

Figure A7.9. Supply Chain Improvements Scenario: Histograms of Farmer Profits
per Hectare, Group 1

2.0T_ _.- - -2 -

Count

 

0.00

1
50000000. 00 10000000000

Profit/Ha, Group 1 (MK)

194

Figure A7.10. Supply Chain Improvements Scenario: Histograms of Farmer Profits
per Hectare, Group 2

 

150-1

 

0 'I
-50000000.00

 

 

1
0.00

5000000000 10000000000

Profit/Ha, Grow 2 (MK)

195

Figure A7 .1 1. Supply Chain Improvements Scenario: Histograms of Farmer Profits

 

per Hectare, Group 3
200
150 -
fl
=
:05 100-
U
501

 

 

 

0.00

5000000000

Profit/Ha, Grow 3 (MK)

10000000000

196

per Hectare, Group 4

“T if i— 77777 V 3

Figure A7.12. Supply Chain Improvements Scenario: Histograms of Farmer Profits

““1

100

Count

50
0

 

1
1
L

6000000000

0.00

5000000000 10000000000

Profit/Ha, Grow 4 (MK)

197

Figure A7.13. Scenario on Quality of Tomatoes Sold in the Market: Histograms of
Farmer Profits per Hectare, Group 1

 

200 5

Count
é

 

 

 

o T—
-5ooooooo.oo 0.00 5000000000 10000000000

ProfitlHa, Grow 1 (MK)

198

Figure A7.14. Scenario on Quality of Tomatoes Sold in the Market: Histograms of
Farmer Profits per Hectare, Group 2

 

150'I

 

 

 

0'1

6000000000

0.00

5000000000

Profit/Ha, Grow 2 (MK)

10000000000

199

Figure A7 .15. Scenario on Quality of Tomatoes Sold in the Market: Histograms of
Farmer Profits per Hectare, Group 3

g

2°°1

150‘

 

 

 

0 '1
6000000000

0.00

5000000000

ProfitIHa, Grow 3 (MK)

10000000000

200

 

Figure A7.16. Scenario on Quality of Tomatoes Sold in the Market: Histograms of
Farmer Profits per Hectare, Group 4

200 "

150‘

 

H
g i
O 100-
O
50 —
0 1 "
60000000. 0.00 5000000000 10000000000
Profit/Ha, Grow 4 (MK)

201

Figure A7.17. Scenario on Low Quality of Tomatoes Sold in the Market: Histograms
of Farmer Profits per Hectare, Group 1

Count

7 1
6000000000 0.00 5000000000 10000000000

Profit/Ha, Grow 1 (MK)

 

202

Figure A7.18. Scenario on Low Quality of Tomatoes Sold in the Market: Histograms
of Farmer Profits per Hectare, Group 2

200T—*r~~——— —— 77‘

Count

 

6000000000 0.00 5000000000

Profit/Ha, Grow 2

100000111000

203

Figure A7.19. Scenario on Low Quality of Tomatoes Sold in the Market: Histograms
of Farmer Profits per Hectare, Group 3

 

Count

 

0.00

5000000000 10000000000

Profit/Ha, Grow 3 (MK)

204

Figure A7.20. Scenario on Low Quality of Tomatoes Sold in the Market: Histograms
of Farmer Profits per Hectare, Group 4

Count

 

0.00

5000000000

Profit/Ha, Grow 4

10000000000

205

BIBLIOGRAPHY

206

 

BIBLIOGRAPHY

Dolan, C. Humphrey, J. and Harris-Pascal (1999). Horticultural Commodity Chains: The
Impact of the UK Market on the Afiican Fresh Vegetable Industry. IDS Working
Paper No. 96. Institute for development Studies. Sussex, UK.

Food Security Research Project (2007). Unban Consumption Survey data.

Hichaambwa, M. and Tschirley, D. (2006). Zambia Horticultural Rapid appraisal:
Understanding the Domestic Value Ch_ains of Fresh Fruits and Vegetables.
Working Paper No. 17. Food Security Research Project.

Humphrey, J. (2007). The supermarket revolution in developing countries: tidal wave or
mgh competitive struggle? Journal of Economic Geography 2007; 7: 433-450.

J affee, S. (1995). The Many Faces of Success: The Development of Kenyan Horticultural
Exports. J affee, S and J. Morton (eds) Marketing Afiica ’3 High Value Foods. The
World Bank . Washington, DC.

Kimenye, L. (1995). Kenya’s Experience Promoting Smallholder Production of Flowers
and Vegetables for European Markets. African Rural and Urban Studies 2 (2-3):
121-141.

Minot, N. and M. Ngigi, (2002). Horticulture Development in Kenya and Ivory Coast. A
Paper Prepared for the IF RI Workshop on “Successes in African Agriculture”,
Lusaka, June 10‘“ —12“‘, 2002.

Minten Bart (2007). The Food Retail Revolution in Poor Countries: Is it Coming or Is It
Over? International Food Policy Research Institute, Discussion paper 00719.

Reardon, T. and CF. Timmer (2006). ‘Transformation of Markets for Agricultural
Output in Developing Countries Since 1950: How Has Thinking Changed?”
Chapter 5 in Handbook of Agricultural Economics, Volume 3. Robert Evenson
and Prabhu Pingali, Eds. Elsevier

Reardon T. and J. A. Berdegue (2002). The rapid rise of Supermarkets in Latin America:
Challenges and Opportunities for Development. Development Policy Review,
Vol. 20, Number 4, PP. 371-388.

Reardon T. and GP. Timmer (2006). Transformation of Markets for Agricultural Output
in DevelopingCountries Since 1950: How Has Thinking Changed? Chapter 5 in
Handbook of Agricultural Economics, Volume 3. Robert Evenson and Prabhu
Pingali, Eds. Elsevier.

207

 

Stevens, C. and J. Kennan (1999). Will Afiica’s Participation in Horticulture Chains
Survive Liberalization? IDS Working Paper No, 106. Institute of Development
Studies. Sussex, UK.

Swemberg K. (1995). Horticultural Exports from Kenya. Horticultural Trade Journal 3:
3-5.

Traill, Bruce (2006). The rapid rise of Supermarkets. Development Policy Review. Vol.
24 Issue 2, Pp333-355.

Tschirley David, Kavoi Mutuku Muendo, and Michael T. Weber (2004). Improving
Kenya’s Domestic Horticulture Production and @etipg System: Current
Competitiveness Forces of Change, and Challenges for the Future (Volume II:
Horticultural Marketing). Tegemeo Institute Of Agricultural Policy and
Development, Working paper 8b, Egerton University.

Tschirley Dave (2007). Suggrmarkets and Beyond: Literature review on Farmer to
Market Linkages in Sub Saharan Africa and Asia. Michigan State University.

Typsa Consulting Engineers and Architects (2004). Inception Report/Consultancy
Services for the Desi g and Supervision of Urban Markets Rehabilitation Project.
Government of Zambia, National Authorizing Office of the European
Development Fund, Ministry of Finance and National Planning.

United Nations (2006). Investment Policy Review. Zambia. United Nations Conference
on Trade and Development.

United States Aid for International Development - USAID (2005). Global Horticultural
Assessment.

Weatherspoon D. and Reardon T. (2003). The Rise of Supermarlgts in Afiica:
Implications for Agrifood Systems and Rural Poor. Development Policy Review,
2003, 21 (3): 333-355.

World Bank (2007). Agricultural Investment Sourcebook: Module 7. Gettinwarkets
Right in the Post-Reform Era in Afiica. World Bank Publication.

208

 

M11111 111111111111111111111111111111111155
3 1293 03063 3808