“it.
$1.? ,_

:1;

, ”my”.
“.3.

,‘un
>7 ' ‘

.4

 

 

 

 

i: ..
”a «Wuhan...
. IL. v_.

.-
$131
..

fl .3 (8 {5.3!
2» r5... 2.44.1.4: . it».
. . a!» 1.3.-
‘4-

v0.54...

. , Qt. .

.a, .

4&wufiww‘
"m 1‘.»

 

L
I.
5

. 1 .n.
‘ . V . :5,
vi. .5va, ,
. L. trawl?“
1.215% :u
.l EC A. ~, 0. , ._ ‘ ;. .
, 4.3.; , «fiétvd. .gfimflwmfifaété. .. .
I ¢ I

. to:
:1} Efiwrflf?
.2... rufinfiudfir.
. , arid
‘Hafifikfia; “4% ..

o

:1 .!
Elonuuw3fl UK).
. , wm.fitb.nn-it ,r.
v .zal.:lr.. rulfh.
. Shun-Jo... ; {14 {.1 Allan In;
in . ‘5 : \
. .ahHflEfl'flih
. . ‘ -
. .—
, r w.
...tt.|_ 5%)... . ‘

nit}...
. ill“? .
“f3

.3“

“ 73:11: ,
”N
d.

.‘ J
I

taalOIBZbél

LIBRARIES
MICHIGAN STATE UNIVERSITY
EAST LANSING, MICH 48824-1048

..-.. .__.

This is to certify that the
dissertation entitled

THREE ESSAYS ON THE RISE OF SUPERMARKETS
AND THEIR IMPACT ON

FRESH FRUITS AND VEGETABLES SUPPLY CHAINS IN KENYA

presented by

DAVID NEVEN

has been accepted towards fulfillment
of the requirements for the

 

PhD. degree in 1 Agicultural Economics

 

 

 

Major Professor’s Signature

waée / fl/i 2007

Date

MSU is an Affirmative Action/Equal Opportunity Institution

--o-.-—-.-.-.-.-.-.-v-.-7.. ..

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.p65-p.15

THREE ESSAYS ON THE RISE OF SUPERMARKETS AND THEIR IMPACT
ON FRESH FRUITS AND VEGETABLES SUPPLY CHAINS IN KENYA

By

David Neven

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

2004

ABSTRACT

THREE ESSAYS ON THE RISE OF SUPERMARKETS AND THEIR IMPACT
ON FRESH FRUITS AND VEGETABLES SUPPLY CHAINS IN KENYA
By

David Neven

Essay 1 The Rise of Kenyan Supermarkets and the Evolution of their Horticultural

Product Procurement Systems

Supermarkets are rapidly penetrating urban food retail in Kenya and spreading well
beyond their initial tiny market niche into the food markets of lower-income groups.
Having penetrated processed and staple food markets much earlier and faster than fresh
foods, they have recently begun to make inroads into the fresh fruits and vegetables
category. The important changes in their procurement systems bring significant
opportunities and challenges for small farmers, and have implications for agricultural

diversification and rural development programmes and policies.

Essay 2 Supermarkets and Consumers in Africa: The Case of Nairobi, Kenya

Supermarkets in urban Kenya have risen from a tiny niche a half decade ago to a fifth of
food retail, spreading well beyond the richer consumers to derive more than a third of
their sales and half of their customers from low income and poor consumers. This essay

explores the patterns and determinants of purchases of the overall food category versus

fresh fruits and vegetables, over Nairobi consumer income strata, for purchases from
supermarkets versus traditional retailers. Implications are drawn for development

programs to help farmers be well-positioned for change in the food markets facing them.

Essay 3 Farm-Level Perspectives on the Impact of Domestic Supermarkets on

Kenya’s Fresh Fruits and Vegetables Supply System.

The rise of supermarkets in Kenya has given rise to a new group of medium-sized farms
managed by well-educated farmers. Focusing on kale, the essay shows that nearly all
supermarket-channel farmers have the capacity to supply larger volumes year round and
have transportation vehicles, an irrigation system, a packing Shed, a cellular phone, and
so on, pointing to the existence of a threshold capital vector which farmers must have in
order to access supermarkets. Especially farm Size and irrigation were found to be
significant determinants of participation in the supermarket channel. Kale suppliers to
supermarkets use more capital intensive production technologies, leading to average
labor and land productivities which are 60-70% higher than in the traditional channel.
Eighty percent of labor consists of hired workers, indicating that these farmers could be
important in alleviating poverty for rural households with little or no land. While most
traditional-channel kale farmers sell to brokers and get a price that lets them break-even
at best, supermarket-channel farmers have a 40% gross profit margin. These margins and
lower market risks in the supermarket channel have resulted in a strong growth dynamic
of supermarket-channel farmers which have doubled the size of their operations over the

last five years.

ACKNOWLEDGEMENTS

It’s been a hard day’s night, as one of life’s greatest philosophers once said.
Neventheless, with a little help from my friends I got by.

My deepest gratitude goes to all the faculty members, staff members and my
fellow students in the Department of Agricultural Economics. 1 want to mention some
specifically. First, I would like to thank my major professor, Dr. Reardon. His humor,
genius and constructive criticism guided me through a sea of discursiveness to the hyper
structures that lead to the good end of the ‘publish or perish’ dictum. I would also like to
thank the members of my dissertation committee, Dr. Peterson, Dr. Raper and Dr. Choi,
both for their interesting classes and their insightful review of this dissertation. I would
further like to thank Dr. Crawford, for sending me the acceptance letter that got the ball
rolling and for his support throughout my doctoral study, Dr. Weber, for his very
appreciated guidance during the first half of my study and for giving me an opportunity
to work in Rwanda, Dr. Clay, for his wise advise and for pointing me in time to that
incredible Spartan basketball team that became national champs in 2000, Dr. Droge of
the Business School, for helping me better understand the academic politics, and my
fellow student Honglin Wang, whose sunny approach to econometrics was very
appreciated.

Thanks also to the people in Kenya who are at the heart of the research presented
here. They were more than a thousand: retailers, consumers, farmers, and so on. A special
thanks also to the project collaborators, especially Dr. Chege of Kenyatta University and

Michael Odera of the Kenya Agricultural Research Institute (KARI). I had a wonderful

iv

experience in Kenya, which is the most beautiful country I have ever visited. Asante
sana!

Without financial assistance my graduate studies and the research presented in
this dissertation would not have been possible. For this I would like to express my
sincerest gratitude to Dr. Weber and the Food Security project, to Dr. Busch and Dr.
Weatherspoon and the Partnerships for Food Industry Development — Fruits and
Vegetables project, to the United States Agency for International Development (U SAID)
which funded these two projects, to Dr. Adesina and the Rockefeller Foundation which
funded the bulk of the field research, to the College of Agriculture and Natural
Resources, for allowing me to really wrap things up, and to the World Bank for
additional fieldwork support. Throughout of course Dr. Reardon has been very
instrumental in obtaining this financial assistance. I would also like to acknowledge the
financial support I received from the Department of Agricultural Economics, the College
of Agriculture and Natural Resources, the Graduate School, the Institute of International
Agriculture (HA) and the International Food and Agribusiness Management Association
(IAMA) for giving me the opportunity to participate in conferences in such exotic
locations as Montreux, Rome and Sidney.

Closer to home, I would like to thank my parents (who gave me every opportunity
and got me well on my way) and my household members Kool, Ringo and little Mel, who
made me realize the relativity of it all when it was needed to realize this most. And last,
but most, I thank my wife Kristin, who because of me had to suffer through five
Michigan winters. She gave me little lovings and that made it all right. Snuit ik hou van

U!

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................ . .................................................... x
INTRODUCTION AND SUMMARY ............................................................................... l
ESSAY 1
THE RISE OF KENYAN SUPERMARKETS AND THE EVOLUTION OF THEIR
FRUITS AND VEGETABLES SUPPLY SYSTEMS ..................................................... 14
1. Introduction ................................................................................................................... 14
2. Data and Methods ......................................................................................................... 15
3. Patterns and Determinants of the Rise of Supermarkets in Kenya ............................... 16
3.1 General Trends ........................................................................................................ 16
3.2 Determinants of Growth ......................................................................................... 19
3.3 A Three-Tiered Supermarket Sector with High Concentration: Emergence of
Uchumi and Nakumatt as Market Leaders .................................................................... 23
3.4 Diffusion Patterns ................................................................................................... 29
4. Supermarket F FV Procurement Systems and Requirements ........................................ 32
4.1 Uchumi: A Decentralised System with a Preferred Supplier Programme .............. 37
4.2 Nakumatt: A Centralised Procurement System Relying on Procurement
Outsourcing through Specialised/Dedicated Wholesalers ............................................ 51
5. Implications for Development Policies and Programs .................................................. 58
5.1 Importance as a Demand-Side Base for Agricultural Diversification .................... 58
5.2 Challenges to Supermarket Procurement Systems in Kenya. ................................. 60
5.3 Impending Challenges to Smallholders .................................................................. 62
5.4 Implications for Development Programmes and Policy ......................................... 65
ESSAY 2
SUPERMARKETS AND CONSUMERS IN AFRICA: THE CASE OF NAIROBI ...... 69
1. Introduction ................................................................................................................... 69
2. Data and Methods ......................................................................................................... 71
3. Supermarkets, Population Dynamics and the Urban Food Market in Kenya ............... 73
4. Empirical Results and Discussion ................................................................................. 78
4.1 Supermarket Customers and Household Income .................................................... 78
4.2 Retail Outlet Choice and Shopping Frequency ....................................................... 87
5. Conclusions and Implications ....................................................................................... 91

vi

ESSAY 3
FARM-LEVEL PERSPECTIVES_ON THE IMPACT OF DOMESTIC
SUPERMARKETSON KENYA’S FRESH FRUITS AND VEGETABLES SUPPLY

SYSTEM ........................................................................................................................... 95
1. Introduction ................................................................................................................... 95
2. Conceptual Approach .................................................................................................... 97
3. Definitions and Data ................................................... , ................................................ 101
4. The Nature of Supermarket-Channel Farmers ............................................................ 106
4.1 Supermarket-Channel Farmers Compared to Traditional-Channel Farmers ........ 106
4.2 The Relative Impact of Export Channel Farmers in the Supermarket Channel 120
4.3 Supermarket-Channel Farmers Categorized ......................................................... 123
4.4 Cases of Smallholder Farmer Groups Supplying Supermarkets ........................... 131
5. The Impact of Supermarkets in the Kale Supply Channel .......................................... 139
5.1 Kale Supply Channels in Kenya ........................................................................... 139
5.2 Determinants of Farmer Participation in the Supermarket Channel ..................... 145
5.3 Supermarket Channel Participation and the Farmer’s Production Technology... 148
5.4 Net Income Effect of Supplying Supermarkets .................................................... 153
5.5 Supermarkets and Farmer Growth Dynamics ....................................................... 158
6. Summary, Conclusions and Recommendations .......................................................... 160

vii

LIST OF TABLES

Table 1: Food and FFV Retailing in Urban Kenya by Type of Outlet, 2003 ................... 18
Table 2: Supermarkets and Hyperrnarkets in Kenya, 2003 .............................................. 24
Table 3: Geographic Distribution of Supermarkets in Kenya, 2003 ................................ 29
Table 4: Uchumi's FFV Supply Composition by Supplier Type, 1997-2008 (%) ............ 40

Table 5: Nakumatt/Mugoya's FFV Supply Composition by Supplier Type, 2003 (%).... 54

Table 6: The Supermarkets' Customer Base by Income Class in Nairobi, 2003 .............. 79

Table 7: Reasons for Retail Outlet Choice in Food by Nairobi Consumers ..................... 81

Table 8: Supermarkets vs. Other Retailers Price Comparison for Processed Foods ........ 82

Table 9: Reasons for Retail Outlet Choice in Fresh Produce by Nairobi Consumers ...... 83

Table 10: Supermarkets vs. Other Retailers Price Comparison for Fresh Produce .......... 84
Table 11: Probit Results Retail Outlet Choice Consumers ............................................... 89
Table 12: Ordinal Multinomial Logit Results Shopping Frequency Consumers ............. 89

Table 13: Land Related Differences (Supermarket vs. Traditional-channel farmers).... 110

Table 14: Labor, Human Capital and Gender Differences (Supermarket vs. Traditional-
channel farmers) ...................................................................................................... l 1 1

viii

Table 15: Physical Capital Differences (Supermarket vs. Traditional-channel farmers) l 14

Table 16: Differences in Diversification, Organizational Capital and Marketing Practices
(Supermarket vs. Traditional-channel farmers) ............................................ 115

Table 17: Determinants of Farmer Adoption of the Supermarket Channel (Probit Results)
................................................................................................................................. 147

Table 18: Heckit Two-Stage Estimates of the Cobb-Douglas Production Functions ..... 151

Table 19: Input Use and Farm Efficiency ....................................................................... 153

Table 20: Gross Margin Analysis Kale Supermarket vs. Traditional-channel farmers.. 156

Table 21: Supply Capacity Supermarket vs. Traditional-channel farmers ..................... 157

Table 22: F armers’ Access Constraints to Key Inputs .................................................... 160

ix

LIST OF FIGURES

Figure 1: Problem Setting ................................................................................................... 2
Figure 2: Growth of Top 5 Supermarket Chains in Kenya, .............................................. 21
Figure 3: Growth of Supermarkets in Broader Macro Context ........................................ 21
Figure 4: Growth of Supermarkets in Sales Floor Space in Kenya, 1976-2002 ............... 24
Figure 5: Growth of Supermarkets in Nairobi vs. Other Urban Centers .......................... 30
Figure 6: Marketing Margins in Kale Supply Chains ..................................................... 142

CBS
COD
COMESA
DFID
EAC

F AO

F C

FDI
FFV
GDP
GoK
HACCP
HCDA
HPHC
KACE
MD
MLE
MoARD
MPV
NGO
OLS
SKU
SMS
SMT
TAV
TCT
USAID

KEY TO ABBREVIATIONS

Central Bureau of Statistics

Cash On Delivery

Common Market for Eastern and Southern Africa
Department For International Development (UK)
East African Community

Food and Agriculture Organization

Factor Cost

Foreign Direct Investment

Fresh Fruits and Vegetables

Gross Domestic Product

Government of Kenya

Hazard Analysis Critical Control Points
Horticultural Crop Development Authority
Horticultural Product Handling Centre

Kenya Agricultural Commodity Exchange
Managing Director

Maximum Likelihood Estimation

Ministry of Agriculture and Rural Development
Marginal Product Value

Non Governmental Organization

Ordinary Least Squares

Stock Keeping Unit

Short Messaging Services

Strategic Management Theory

Traditional African Vegetables

Transaction Cost Theory

United States Agency for International Development

xi

INTRODUCTION AND SUMMARY

The Rise of Supermarkets in Developing Countries: General Problem Context

Changes in end-consumer demand (population growth, urbanization, rising incomes),
policy shifts (trade and market liberalization) and globalization/foreign direct investment
(F DI) have spurred a structural transformation process from traditional to modern in the
agri-food systems of developing countries. At the forefront of this modernization process,
as key agents of change, are the supermarkets, which have grown from niche players in
the high income segment to the dominant actors in the food system'.

The growth of supermarkets has downstream and upstream ripple effects. For
consumers, on the one hand, supermarkets offer a greater variety of food products, value-
added food products, more guarantees regarding the quality and safety of food products
and generally lower prices for processed food. On the other hand supermarkets take
market share away from traditional retailers (such as kiosks, over-the-counter Shops,
markets stalls, and street hawkers), lowering their margins or displacing them, and thus
may reduce the set of retail outlets from which consumers can choose.

From the perspective of their suppliers, supermarkets are very demanding
customers. They require higher and more consistent quality, consistent year round
delivery, larger volumes per supplier, more stringent payment terms (e.g., a lS-day

payment period rather than cash-on-delivery), and so on. Suppliers who are most

 

l ‘Supermarkets’ is a term used broadly here for all self-service retail outlets meeting minimum Size criteria
(150m2 in the case of Kenya) and with food lines representing an important percentage of sales (>50%).

successful in meeting these criteria face great grth opportunities, but those who cannot
make the grade are likely to be dropped.

As the market share of supermarkets in food retailing grows, the effects of this
vendor rationalization process will not be marginal. This has raised considerable concern
on how the most vulnerable agents in the food system adapt. For smallholder farmers
there is a risk that they may not be able to adapt to the superrnarkets’ new procurement
system, leading to exclusion from a potentially more profitable or less risky market.
Alternatively, some small farms may adapt, through collective action, investments and
learning, taking full advantage of the grth opportunity offered by supermarkets. Figure

1 depicts the problem setting.

Figure 1: Problem Setting

 

 

 

 

 

 

Market liberalization Rise of Changing ' Downstream and
Trade liberalization supermarkets structure of upstream effects on
Population growth and other the supply —+ (smallholder) farmers
Urbanization modern retail chain and (poor) urban
Globalization/FDI formats consumers

 

 

 

 

 

 

 

 

 

 

 

 

Supermarkets in Kenya: General Research Approach

The research presented here focuses on the impact of the rise of domestic supermarkets
on urban consumers of and supply chains for fresh fruits and vegetables (FFV) in Kenya.
Kenya provides an interesting case study because it demonstrates how even in a low
income country in Sub-Sahara Africa, supermarkets are growing with all the

irrepressibility of an idea whose time has come. In Kenya, we show here how

supermarkets have grown from a tiny niche at the start of the 19905 to 20% of the urban
food retail sector in 2003. Furthermore, supermarket development in Kenya is currently
in an early, formative stage where changes are taking place fast. This study thus allows us
to capture these fleeting moments and provide a benchmark for future studies on urban
food retailing in Kenya. We focused on FFV because it is an important sub-sector in the
country, both from the consumer perspective (important part of the diet, the food
expenditure) and the farmer perspective (great income potential for nearly 2 million small
scale producers in Kenya). Within this context, this study aims to help answer the
following three strategic questions, each at a different level of the supply chain: (1) at the
retailer level: ‘What is the importance of supermarkets in food and FF V retailing and
what is the nature of the supermarkets’ F F V procurement system? ’; (2) at the consumer
level: ‘What is the effect of the development of the supermarket sector on the (urban)
consumers ’ food and FF V purchasing behavior? ’; and (3) at the farmer level: ‘What is
the eflect of the development of the supermarket sector on (F F 10 producers’ behavior
and net income? ’.

The approach to the research questions is framed in a broad, dynamic
interpretation of the classical Structure-Conduct-Performance model, where the structure
of the supply system determines the behavior of economic agents (be they retailers,
consumers or farmers) and the behavior of agents in turn results in a new structure of the
supply system. As the system evolves, there will be differential effects in terms of the
performance of the different agents. Within this general framework, standard adoption
models were used to analyze the choices agents make, i.e., choice is hypothesized to be

determined by the incentives and capacities of agents which in turn are determined by

input and output prices, risk factors, quasi-fixed capital and shift factors. The theories
from institutional economics, especially with regard to the role of transaction costs, are
brought into the model to refine the interpretation of the independent variables. A variety
of analytical methods was used, including case studies, analysis of descriptive statistics,
econometric estimations of adoption and technology functions and farm-level, product
specific gross margin analysis.

The data for this study were collected over 13 months of field research in Kenya
(J anuary-November 2003, April 2004, and August 2004). Given the absence of secondary
data on retail in general, and on supermarkets in particular, in Kenya, the bulk of the
research effort focused on primary data collection. A total of 1,174 respondents were
interviewed. These interviews came in various formats and included key expert

interviews, focus group discussions and surveys.

Retailer, Consumer and Farmer Perspectives on the Rise of Supermarkets in Kenya

This study takes on the format of a collection of three closely related essays. Each essay
zooms in at a different level of the supermarket driven supply chain, each addressing one
of the three strategic questions listed above. While the three essays are linked,
complementing each other and jointly providing a complete picture of the supply system

from farmer to consumer, they are written so that they can be read as individual essays.

Essay 1: The Rise of Kenyan Supermarkets and the Evolution of their Fruits and
Vegetables Supply Systems

Research Questions

Essay 1 goes to the heart of the phenomenon under study, the importance of
supermarkets, the drivers behind and patterns of their growth and the nature of their
produce procurement system. The Specific descriptive and research questions are: (1)
what is the importance of supermarkets in terms of market Share in food and FFV, growth
rate and geographic Spread? (2) How does growth of a supermarket chain determine: (i)
the choice of the FFV supplier (in terms of their Size and place in the supply chain); and
(ii) the use of contracts, quality and food safety standards and centralization of

procurement?

Data and Methods

The data collection effort consisted of (1) a supermarket survey (210 respondents, usually
the branch manager, Spread over 79 urban areas); (2) a similar survey of traditional
retailers such as kiosks, over-the-counter shops and market stalls (250 respondents); and
(3) key informant interviews (including interviews with the successive Managing
Directors, since 1986, of Kenya’s leading supermarket chain, Uchumi). Descriptive
analysis of this data set, complemented with secondary data, allowed for a reconstruction
of the path of supermarket development in Kenya and the identification of the key drivers
behind it. The nature of the F FV procurement system is analyzed through case-studies of
the two leading supermarket chains in Kenya, Uchumi and Nakumatt, who jointly

represent 90% of the FFV sold through supermarkets in Kenya.

Key Findings

(1) Growing at rate of 18% per year, supermarkets in Kenya have grown from a niche in
the mid 19908 to 20% of the urban food market and 4% of the urban FF V market in 2003.
(2) Supermarkets have Spread from the capital to intermediate and small towns, with 44%
of supermarket sales and 58% of supermarket stores located outside of Nairobi in 2003.
(3) The leading supermarket chains in Kenya are evolving, with a time-lag but at a
comparable speed, to the same FFV procurement system that has become the dominant
format for supermarkets in middle-income developing countries: (i) while the share of
brokers (and the smallholders they buy from) in the vegetable supply of Kenya’s leading
supermarket chain (Uchumi) has gone down from 70% to 50% (and is expected to go
further down to only 10% by 2008), the share of direct supplies by medium and large
farms has gone up from 15% to 40% (and is expected to go up to 65% by 2008) and the
share of smallholders as direct suppliers remains stable around 10-15%; (ii) both leading
supermarket chains have moved towards the use of more centralized procurement
systems for FFV; and (iii) there is an incipient use of standards for food quality and food
safety, mostly linked to contracts, the use of which is expected to increase from the

current 5% of FFV suppliers to 70% over the next 5 years.

Essay 2: Supermarkets and Consumers in Africa: The Case of Nairobi, Kenya

Research Questions

Essay 2 looks downstream from the retailer to the end-consumer. The supermarkets’ gain
in market share can only be explained by a concurrent shift in consumer behavior. The
specific descriptive and research questions are: (1) How important are the different
income-categories of consumers as a percentage of the customer base and sales of
supermarkets for food in general and F F V in particular?; and (2) What are the socio-
economic determinants of the consumers’ retail outlet choice (supermarkets vs. other

retail outlets) and shopping frequency decisions (at supermarkets) for food and FFV?

Data and Methods

The data for were collected through eight consumer focus groups (54 participants) and a
consumer survey (445 respondents) in Nairobi. The consumer focus groups provided an
initial understanding of the behavior of consumers and aided in the design of a Short
consumer survey instrument. The data from this survey were then used to: (i) describe the
nature and motivations of consumers shopping in supermarkets versus traditional retail
outlets across product groups and income classes; and (ii) test, in limited dependent
variable regressions, the importance of different socio-economic variables (including

income) in explaining retail outlet choice and shopping frequency.

Key Findings

(1) The poor are far more important in the development of supermarkets than is generally
believed: in 2003 supermarkets derive more than a third of their sales and more than half
of their customers from poor and low income consumers.

(2) In general, supermarkets have penetrated much further and faster into the consumer
market for processed foods as compared to FF V. This is mainly because Nairobi
supermarkets sell processed food much cheaper than do traditional shops. By contrast,
supermarkets sell F F V at a higher price (on average) and are not located as close to
consumers as are FF V shops. However, there are three factors that suggest that Kenyan
urban consumers will shop more from supermarkets in the future for FFV: (a) younger
consumers buy F F V more frequently from supermarkets; (b) supermarkets are
increasingly price-competitive with many categories of traditional retailers for high-
volume (commodity) vegetables like kale and tomatoes; and (c) supermarkets are

developing smaller retail formats which can be located closer to consumers.

Essay 3: Farm-level Perspectives on the Impact of Supermarkets in Kenya’s Fresh
Fruits and Vegetables Supply System.

Research Questions

Essay 3 takes the upstream, farm-level perspective on the growth of supermarkets. The

research question here are: (1) what are the determinants of the farmer’s decision

regarding channel choice (supermarkets versus traditional buyers) in terms of their

capacities (farm size, use of technology, use of hired labor, education and gender of the

farmer, and so on); (2) do farmers supplying supermarkets and farmers selling to

traditional supply chain agents use different production technologies (in terms of the
relative amounts of labor, land and capital used); and (3) do farmers supplying
supermarkets benefit from this market access in terms .of net incomes and/or demand

reliability?

Data and Methods

Two farmer surveys, mainly focusing on kale, tomato and banana (with 63 of the
interviewed farmers supplying supermarkets and 103 supplying traditional markets),
complemented with semi-structured interviews with supply chain intermediaries and
input suppliers, were conducted to address these questions. The first survey takes a
broader perspective, focusing mostly on kale, tomatoes and bananas (the two highest
volume vegetables and the highest volume fruit in Kenya respectively). The descriptive
data from this survey were used to address question (1) above. The second farmer survey,
this time focusing on kale only and including detailed questions on input use and yields,
addresses all three questions. This part of the analysis consequently consists of three
steps. First, a probit implementation model is used to test channel adoption by farmers in
terms of their capacities. Second, the essay tests the production technology differences
between supermarket-channel farmers and traditional-channel farmers by comparing their
marginal and average product values for various input factors in kale production. Third,
gross margin analysis is used to assess the income benefit farmers derive from being able
to sell to supermarkets. Qualitative analysis and descriptive statistics are further used to
assess: (i) the role of market risk reduction as a benefit to the farmer; and (ii) the presence

of a growth dynamic at the farmer level in the supermarket channel.

Key Findings

(1) The growth of supermarkets has stimulated the rise of a new class of F FV farms in
Kenya. These are mostly recently established, medium-sized farms which, in terms of
physical and human capacities, differ starkly from the smallholder farmers that represent
the bulk of the F FV production in Kenya. Especially farm size and the presence of a
modern irrigation system were found to be critical determinants of participation in the
supermarket channel. This stark difference points to the presence of a threshold capital
vector at the entrance of the supermarket channel.

(2) Supermarket-channel farmers have adopted more capital intensive production
technologies which resulted in their average land and labor productivity being 60-70%
higher than those of traditional-channel farmers. While they use 20% less labor per acre
than traditional-channel farmers, supermarket-channel farmers rely heavily on hired labor
(80% of the workforce) and could thus become important in proving jobs for rural
households with little or no land. Not withstanding differences in the capital intensity of
their production technologies, the two types of farmers have a comparable production
cost per unit of output.

(3) While most traditional kale farmers sell to brokers and get a price that lets them
break-even at best, supermarket-channel farmers have a 40% gross profit margin. These
margins and lower market risks in the supermarkets channel has resulted in a strong
growth dynamic for supermarket-channel farmers which have doubled the size of their

operations over the last five years.

10

Implications for Development Programs

(1) The growing market share of supermarkets implies that they are becoming
increasingly important as transformers of the food markets facing small farmers, and thus
as agents to include in development programs aimed at improving market access to the

small scale farmers.

(2) Small farmers, as individual suppliers, are one of the less attractive options for
supermarkets to turn to as their FFV sales increase, because of the former’s inability to
deliver year-round and in the volumes and at the quality level required by supermarkets.
Development programs aimed at assisting smallholder farmers to access the supermarket
sector would have to make sure farmers meet the whole vector of supermarket
requirements. As this implies investment and working capital requirements which mostly
exceed the capacities of risk-averse smallholder farmers, development programs need not
only organize farmers in groups but also look for or develop new creative ways of
keeping capital requirements low (e.g., through factoring) and increase access to
affordable credit. Given the wide scope of the criteria that need to be met, development

programs will likely have to include multiple partners, specialized in different fields.

(3) Supermarkets prefer to work with a selected list of proven suppliers, who grow along
with them. As these farmers are already selected in the early stages of the development of
the supermarket’s F FV procurement system, there is a degree of urgency in helping

smallholder farmers to sell to supermarkets and build a long term trading relationship.

11

(4) Supermarket-induced supply chain dynamics (such as increased investments in
irrigation, communication, transportation) result from an increase in scale and from an
increased market reliability (long term relationships). Development programs aimed at
creating similar dynamics in the traditional channel (which will dominate the FF V market
for the foreseeable future in Kenya) should therefore look at investment stimulating scale
increases at the wholesale and retailer level (e.g., small Shops aggregating to form

procurement clubs that can then contract directly from groups of smallholder farmers).

(5) In order to avoid the risk of an over-reliance on a single buyer and because
supermarkets buy only the highest quality grades, farmers Should look at supermarkets as

one element in a portfolio of buyers.

(6) Supermarkets, as they centralize their F FV operations over a wider regional store
network and as they increase their scale of operations, will stimulate regional market
integration and production specialization, with the best suppliers for particular products
increasingly found farther away from the area of production (this includes increased
imports and exports as chains move into new countries). When selecting products and
developing marketing plans for smallholder farmers, development programs should take

these dynamics into account (both as opportunities and threats).

(7) Assistance programs focused on agricultural diversification can help small farmers

with the necessary training and equipment, such as transport and cold chains, to produce

12

those perishables for which these farmers are likely to be competitive in the supermarket
market-channel. This assistance Should be carried out in a way that walks the tightrope
between alleviating poverty as much as possible, and inducing distortions in market or
investment incentives. We Show in this study that these products are mainly the highly
perishable products such as leafy greens (e.g., traditional African vegetables) in which a
correctly equipped smallholder can have an advantage. Assistance programs are likely to
find further promising new market opportunities amongst the wide set of produce items
for which supermarkets are eager to shift away from the current traditional brokers to
more direct supplies by farmers (e.g., potatoes, carrots). In some cases this will imply
producers from different regions to cooperate with each other in order to address the
year-round delivery requirement of supermarkets currently resolved by brokers (e.g., for
mangoes). By contrast, we expect rapidly increasing cost and quality competition for

small farmers competing in this channel in bulk products such as bananas and tomatoes.

13

ESSAY 1

THE RISE OF KENYAN SUPERMARKETS AND THE EVOLUTION OF THEIR
FRUITS AND VEGETABLES SUPPLY SYSTEMS

1. Introduction

Recent research has drawn attention to the rapid growth of supermarkets and the resulting
structural transformation in the agrifood systems of developing countries (for example,
Reardon and Berdegué (2002) for Latin America, and Weatherspoon and Reardon (2003)
for Africa, focusing mainly on South Africa). In this essay we demonstrate that the
supermarket sector is also developing rapidly in Kenya, a poor country (with a per capita
GDP of roughly a dollar a day) where supermarket growth is endogenous and indigenous
— without the heavy influence of foreign direct investment (FDI) by global retail chains
that plays such an important role in supermarket trends in Asia and Latin America.
Supermarkets in Kenya have grown from a tiny niche market only seven years ago to
20% of urban food retail today, and are rising fast. At issue is how this phenomenon is
affecting food markets facing farmers, via changes in supermarket procurement systems.
We focus on the fresh fruits and vegetables (FFV) category, because government
and donors have high hopes of this being a motor of agricultural diversification for small
farmers, given FFV’S assumed lack of scale economies, and because the relationship
between supermarkets and F F V producers tends to have fewer intermediate steps, in
particular in processing; procurement system changes are therefore expected to change
more directly the market facing small farmers. Moreover, FF V in Kenya is a large sector,

producing 3.2 million tons of fruits and vegetables with a rural-market value in 2002 of

14

US$354m. (3% of GDP), and the majority of horticultural producers are smallholder
farmers (providing 70% of the marketable FFV) who derive a large share of their cash
income from horticulture (MoARD, 2002; Karnau, 2000). Prior work on the supermarket
channel for FFV sales has focused on exports (Jaffee, 1994) including specifically to UK
supermarkets (Dolan and Humphrey 2000), but the literature is bereft of studies on F F V
sales to supermarkets in Kenya. Yet the great majority of Kenyan FFV goes to the
domestic market (90% of the volume and 70% of the sales), and, as we Show below,
supermarkets in Kenya already buy half as much FFV as are exported (in volume terms).
Finally, domestic FFV marketing studies, such as Dijkstra (1997, 1999a, b, 2001), do not
cover local supermarkets.

The essay proceeds as follows. Section 2 discusses the collection of primary field-
survey data on which the study is based. Section 3 presents data on trends in the
development of the supermarket sector in Kenya. Section 4 analyses the Kenyan
supermarket chains’ procurement system changes for FFV and discusses the implications

for farmers of these changes. Section 5 concludes with policy and program implications.

2. Data and Methods

Given the absence of supermarket (or retailer) associations and industry publications,
secondary data on supermarkets in Kenya are scarce (basically limited to supermarket
annual reports and articles in the popular press).

Hence this essay is based on primary, field-survey data collected by the author

during March to November 2003 and April 2004. (i) In-depth interviews were conducted

15

with the executive managers and/or F FV managers of the top five supermarket chains,
plus visits to their key FFV wholesale suppliers. (ii) A short survey was administered in
79 of the 87 urban areas with over 25,000 population (according to the 1999 population
census, CBS, 2002b), where all supermarket stores found in these urban areas were
visited and either a manager interviewed and/or observational data collected; in total, 210
valid interviews in this category were conducted. (iii) A similar survey instrument was
used to interview 250 non-supermarket retailers (smaller self-service groceries and
traditional retailers such as greengrocers, open air market stalls, kiosks, street hawkers
and over-the-counter Shops). (iv) A survey of 450 households in Nairobi, focused on their
shopping habits. (v) A varied set of additional data was gathered, including interviews
with government officials, key industry experts in the retail and agri-food industry

sectors, and other private sector firms, and NGOs involved in the FF V industry.

3. Patterns and Determinants of the Rise of Supermarkets in Kenya

3.1 General Trends

We use the definition used by ACNielsen-Kenya for supermarkets, as ‘self-service stores
handling predominantly food and drug fast moving consumer goods (FMCG) with at
least 150m2 (1,625 sq.ft) of floor space’. We defined ‘hypermarkets’ as having at least 15
times that floor Space (2,250m2). Using these definitions, as of 2003, there were 225

large-format stores in Kenya — 209 supermarkets and 16 hypermarkets.2 In the text, for

 

2 We do not include in our analysis the estimated 900 to 1,400 mini-supermarkets and convenience stores,
also non-tradrtronal retail formats, that have emerged in recent years in urban Kenya.

16

simplicity, we use the term ‘supermarkets’ to refer to both formats, distinguishing only
where needed. Note that, while the cut-off point for the definition of a supermarket is
relatively low by international standards (where the threshold is usually 300-400 m2), it is
judged in relation to the traditional store Size by the ACNielsen definition. However, it
should be noted that the average Size of a supermarket in Kenya is 9,900 sq. ft. Moreover,
most (80%) of the stores are part of a chain of supermarkets. Thus, small independent
supermarkets are a small share of the sector.

Supermarkets have taken market share away from traditional food retailers such
as kiosks, greengrocers, over-the-counter Shops, market stalls and street hawkers (Table
1). The supermarket sector had an estimated turnover of US$520m. in 2002. Supermarket
food sales represent roughly 70% of total sales — US$365m. Using triangulation from
macro-data, our consumer and retail surveys and secondary data, we estimate the total
size of the urban food retail market at US$19 billion. The market Share of supermarkets
in tetal urban food retailing is therefore roughly 20%. (This is similar to Indonesia now or
Mexico five years ago.) Sub-supermarket-size self-service shops represent another 17%

of the market, and the rest, 63%, is the Share of traditional retailers.

17

Table 1: Food and FFV Retailing in Urban Kenya by Type of Outlet, 2003

 

 

Food Sales FFV Sales
Type of outlet No. of USS Market USS Market
outlets m. share m. share
, (%) CA)
Kiosks, OTC shops, greengrocers 27,000 600 32 131 36
Market stalls and street hawkers Na 590 31 211 58
Supermarkets and hypermarkets 225 365 20 15 4
Smaller self-service Shops 1,200 310 17 7 2
All types Na 1,865 100 364 100

 

Note: the exchange rate used was lUS$=Ksh76.9.

Sources: author’s estimations based on authors’ retailer and consumer surveys and secondary
data (AC Nielsen, 2002; CBS, 2002a, b, c; World Bank, 2004; World Gazetteer, 2004; UN
Population Division, 2003; UN Habitat, 2002; UN Statistics Division, 2003).

Over the next 5 to 10 years, the supermarket sector is expected to grow at a rate Similar to
the current growth rate, i.e. about 18% per year since 19953. At this growth rate, and
assuming a market growth (urban population growth) of 4.7% (UN Population Division,
2003), supermarkets will double their market share by 2009 and become the dominant
urban food retailers by 2011. This may happen even earlier if global, Asian or South
African chains include F D1 in Kenya in their strategies in the near future, which is
perhaps a reasonable assumption.

The supermarkets’ share in FFV retail has lagged behind their overall penetration
of food retail (as is common in other parts of the world, for example in Latin America,
see Reardon and Berdegué, 2002), but the trends are parallel. We estimate that out of
US$364m. worth of FFV marketed to the urban consumer in Kenya in 2002,
supermarkets represent US$l4m. or 4% (i.e., an estimated 35,000 MT). As expected, the

share is higher in large cities; in Nairobi the market share is 6-7%.

 

3 We expect this because the drivers of growth (urbanization, competition, trade and market liberalization;
see section 3.2) are expected to continue.

18

However, it is interesting to compare supermarket sales of F FV (circa 35,000
tons) with exports of FFV — 67,000 tons (MoARD, 2002). While research and policy
attention has focused either on the export market or the traditional market, this third
market - the supermarket market — represents already 50% of exports in terms of
volumes. Because of the higher quality, higher value-added and different product
composition (e.g., shelf-ready pre-packs of Asian vegetables for UK supermarkets), the
same cannot yet be set about the value. While supermarkets sold an estimated $14m.
worth of fresh fruits and vegetables in 2002, the exported FOB value was $160m. In
terms of farm value, the gap closes with supermarkets representing 89m. and exports
$45m. as exporters have higher marketing margins than supermarkets (roughly 70% vs.

40%).

3.2 Determinants of Growth

While the supermarket is not new to Kenya (the first supermarkets arose in the 19605),
their rapid grth is a very recent phenomenon, having taken off since the mid-19905.
This growth of the supermarket sector in Kenya has mainly been driven by three factors
that coincide with some of those important in supermarket diffusion in other developing
countries.

First, there has been rapid urbanization: the urban population made up 13% of the
total in 1975, and 36% in 2000, and is expected to surpass the rural population by 2013
(UN Population Division 2003). Based on these UN statistics, the average urban
population growth rate (4.7%) is triple that of the overall population (1.6%), as the rural

population has started to decline even in absolute numbers. The populations of

19

intermediate cities like Nakuru and Eldoret doubled between 1989 and 2002 (World
Gazetteer, 2004). Moreover, new registrations for buses and mini-buses, mostly
consisting of public service vehicles providing intra and inter-city transportation,
quintupled between 1993 and 2002 (CBS, 2002b, 2003).

Second, trade and domestic market liberalization started in 1993. Figure 1 depicts
the growth of Kenya’s current top five supermarket chains over the period 1975 (when
Uchumi, Kenya’s largest supermarket chain, was established) to 2003. It is a kinked
curve with growth really taking off in 1995, i.e., after the 1993 policy changes were
starting to have an effect. The 1993 economic reforms, including liberalization and
stabilization policies, had several important effects for supermarkets. (i) Import licensing
removal (more imports) and market liberalization (more domestic competitors) led to a
dramatic increase in product variety and shifted the retail market from a sellers’ to a
buyers’ market in which retailers had to fight for the consumers’ shilling vote. The direct
consequence of the increased product variety in the marketplace is that it favored the
bigger stores (supermarkets) able to stock a wider assortment of products (economies of
scope). (ii) Price liberalization also played into the hands of the supermarkets because it
facilitated, in Kenya, the low margin-high turnover strategy that has been the core of
most of the supermarket growth around the world. (iii) There was a mild and short-lived
recovery of the economy in 1995/6 (with annual real GDP growth in the 4% range) which
gave consumers the buying power to try all these new products that supermarkets were
marketing to them. However, note that Kenya’s economy deteriorated afterwards, ending
the 19905 with no growth. With a downward trend in real GDP per capita over the last

half-decade (CBS, 2002a), there is no indication of a general rise in consumer incomes in

20

Kenya. Yet, as Figure 2 Shows, the supermarkets’ fast growth has persisted even in these
slack economic times. This is probably due to their being able to offer low-priced

processed foods and staples, appealing to lower middle-income consumer groups.

Figure 2: Growth of Top 5 Supermarket Chains in Kenya,
1975-2003

 

 

 

 

thunborolshuns
8888

K

 

 

 

VVVV I I T I V V Y Y T T V V I V I T I Y I I V Y I 1

1975 1977 19791981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

0

Year

 

 

 

Source: author’s national supermarkets survey.

Figure 3: Growth of Supermarkets in Broader Macro Context

 

 

 

 

 

 

 

 

 

 

in Kenya, 1985-2003

A 1.000
8
E +Supenmrlets'SalesFloor
3 Space
5 +UbanPbpulation
g —o—rethP
0
; —rethP/capita
E
a 50 WW.

1985 1988 1991 1994 1997 2000

year

 

 

 

Sources: sales floor space is based on author’s supermarket survey;
population data from Heston 2002; GDP data are from Econstats, 2004.

Third, while Kenya’s supermarket revolution is unlike that of most developing countries

outside of Africa that had massive inflows of retail F DI after investment liberalization in

21

the mid/late 19905 (but similar in the following respect to South Africa, see
Weatherspoon and Reardon, 2003), Kenya’s supermarket sector growth has been almost
completely indigenous and endogenous. However, inter-chain competition in domestic
investment has been very vigorous, playing the role of foreign versus domestic chains’
investment competition seen elsewhere. Of course, ‘if’ and when outside chains do enter,
the intensity of the investment rivalry will increase. The entry of foreign retailers afier
domestic supermarkets reach a critical point in their growth is what has been observed in
other developing countries (e.g., in Argentina; Ghezan et a1. 2002) 4. These foreign chains
bring with them high levels of operational efficiency and spark a price war which forces
all supermarkets to speed up the already initiated process of driving down costs and
pushing into new, less competitive but also less rich markets (smaller towns, lower
income neighborhoods).

Before 1993, the main supermarkets grew within the boundaries of a single town,
leading to local fiefdoms. Uchumi broke this pattern in 1993 by building its first store
outside Nairobi, in Nakuru, starting a national-level competition that has built-in
crescendo. Most notably, the rivalry between leading chains Uchumi and Nakumatt
became an important growth driver; a new strategy by one chain forces imitation and/or a
counter strategy by its competitor. For example, Nakumatt’s introduction of large-forrnat
stores in 1995 led to the introduction of hyperrnarkets by Uchumi in 1997. Uchumi’s
subsequent introduction of a fully-fledged FFV department in that year (coinciding with

its first hypermarket in Nairobi) was followed by Nakumatt in 2001. Over time, this

 

4 The interest of South African supermarket chains to enter the Kenya markets dates back to at least 1994
when Pick‘n Pay offered to buy Uchumi (author’s interview with Suresh Shah, MD Uchumi 1986-99, 27
March 2003).

22

competition has moved from such conspicuous differentiation Strategies to the more

subtle price-based competition implying economies of both scale and scope.

3.3 A Three-Tiered Supermarket Sector with High Concentration: Emergence of
Uchumi and Nakumatt as Market Leaders

Kenya’s supermarket sector has three tiers (Table 2). The first tier consists of the two
clear market leaders and also the leading FF V retailers, Uchumi Supermarkets and
Nakumatt, which together control nearly 50% of the supermarket sector. The second tier
consists of Tusker, Ukwala, and Metro Cash ‘11 Carry chains. These top five chains have
28% of the large-format stores in Kenya and 60% of the sales, indicating a concentrated
sector (Similar to Latin American and European levels of concentration). Figure 3 shows
that since 1996, in terms of store size (a proxy for sales), the top five have surpassed the
other supermarkets and are growing at a faster pace, increasing their dominance over
time. The supermarkets in the third tier consist of small chains (of vwhich there are about
40) and independent (single-store) supermarkets. There is only one publicly traded
Kenyan supermarket (Uchumi) and one foreign-owned supermarket of significance
(Metro).

While selling FFV in supermarkets is not a rarity, given that overall 4 out of 10
supermarkets sell them to various degrees, it is highly concentrated in the hands of two
leading chains. Of the Kshl billion worth of FFV sold through supermarkets, Uchumi

and Nakumatt represent roughly 70% and 20% respectively.

23

Table 2: Supermarkets and Hypermarkets in Kenya, 2003

 

Average Stores
Super- Hyper- Total store Market

 

 

 

 

 

 

 

 

 

sellin .
Company markets markets stores size 3?"; FFVg Ownership
(sq. ft) (‘4 9‘ "I 1%)
Market leaders
Uchumi 23 4 27 16,400 26 100 Public Kenyan
Nakumatt 4 8 12 48,700 20 100 Asian Kenyan
Second tier
Wsker 7 l 8 14300 6 25 African Kenyan
Ukwala 9 0 9 9,100 4 0 Asian Kenyan
Metro 2 l 3 28,400 4 0 South African
Third tier
54% Afric. Kenyan
Small chains 71 1 72 6,200 24 40 44% Asian Kenyan
2% Other
51% Afric. Kenyan
Independents 93 l 94 4,200 16 22 45% Asian Kenyan
4% Other
All supermarkets
PM
o rc. enyan
All 209 16 225 9,900 100 41 42% Asian Kenyan
4% Other

 

Note: market shares refer to total sales and are based on sales estimated from store sizes, interviews with
upper-level supermarket managers, the supermarket survey and Uchumi’s annual reports.
Source: author’s national supermarket survey.

Figure 4: Growth of Supermarkets in Sales Floor Space in Kenya, 1976-2002

 

1.400.000 - ~
1.200.000
1.000.000
800.000
600.000 «
400.000 - ,
200.000

0 " '
1976 1979 1990 1992 1994 1986 1999 1990 1992 1994 1996 1999 2000 2002

 

 

 

 

 

 

 

 

 

 

Sq. Ft

 

 

 

 

 

 

 

 

Year

 

 

 

F_ Top 5 Supermarkets Chains ------- Other Supermarkets

 

 

Source: Authors’ national supermarkets survey.

Uchumi

Uchumi Supermarkets Ltd was established as a wholly govemment-owned company

from three existing stores in Nairobi in 1975. In 1992, as part of the govemment’s

24

ongoing structural adjustment programmes, Uchumi became a public company traded on
the Nairobi Stock Exchange. However, its three main shareholders, who jointly owned
51% of the stock in 2003, are the original ones, namely, ICDC, wholly govemment-
owned, and KWA and ICDCI, both of which have the government as a key shareholder
(Uchumi, 2003). This means that the government has a strong presence on the board of
Uchumi which has to approve major strategic decisions such as investments.

A new phase in the company’s history began in 2001 when Uchumi embarked on
an ambitious five-year expansion plan, key features of which include an increase in the
number of stores from 17 to 50, an expansion into the regional market (Uganda,
Tanzania), the construction of a new distribution centre in Nairobi and the installation of
a satellite-based IT system linking all its stores to this new centre. Good progress was
made at first: 10 new stores (including two hyperrnarkets in Nairobi and one in Kampala,
Uganda) had opened by December 2002 (doubling its total floor space from 215,000 to
450,000 sq. ft over this two-year period), the new distribution centre was finished by
2003 and the new IT system started in September 2003. However, a combination of high-
interest Short-term financing tools and investments outpacing sales growth led to the first
annual 1055 in the company’s history (for the fiscal year ending 30 June 2003). The
subsequent backlash from wonied investors led to the decision to put the 5-year
expansion plan on hold for a year.

Uchumi took $135m. in sales in 2003 (Uchumi, 2003), about 1% of Kenyan GDP,
more or less the same as Wal-Mart/U S sales are of US GDP. Uchumi today is the market
leader with 27 branches in seven urban areas throughout the country (Nairobi, Mombasa,

Nakuru, Eldoret, Meru, Karatina, Kisii) and one branch (a hypermarket and the

25

company’s flagship store) in Kampala, Uganda. These 28 branches include various store
formats, reflecting the fact that Uchumi targets customers from all socio-economic
classes. Its 4 hyperrnarkets with large parking areas along the main entry/exit roads in
Nairobi mainly attract high- and middle-income consumers. Its smaller ‘neighborhood’
stores in the city’s residential areas (‘estates’) mostly target middle-income consumers.
Its city centre stores near busy bus stages (5,000-20,000 sq. ft) mostly attract the middle-
to low-income consumers.

In 1997, Uchumi was the first major chain in Kenya to introduce FF V items in its
stores (as part of an overall strategy of building up ‘fresh’ categories: dairy, meats, bread,
FFV). Starting out with some trial sales, FFV sales are now over Ksh50m. per month.
Today, all Uchumi branches have a FFV section, although their size and assortment vary
with the customer-base, with the most developed sections found in the hypennarkets
where they can take up 7% of the sales floor space and offer over 300 stock keeping units
(SKUs) from American pink sweet potatoes to zebra yellow melons (although only 100
or so are available at any one time). The FFV section in hyperrnarkets also includes
value-added (and increasingly branded) products such as a line of organic products or
pre-cut vegetable packs. In stores targeting lower-income consumers, the FFV sections
are smaller, carry fewer SKUs and focus on the staple F FV items of the poor (such as
kale, spinach, cabbage). Notwithstanding this broad customer targeting, FF V sales at
Uchumi are still very much focused on Nairobi (90% of sales) and its hyperrnarkets (60%

of sales).

26

Nakumatt

Privately-owned Nakumatt was established in Nakuru in 1985 (then still Nakuru
Mattresses) and remained a small operation until it moved to Nairobi in 1992. By
assertively opening stores in key locations and by expanding existing ones, Nakumatt
grew into a major supermarket chain with 12 large-size branches in Kenya’s three main
urban centres (9 in Nairobi, 2 in Mombasa, l in Kisumu). Nakumatt’s growth is
facilitated by being part of a network of companies with cross-shareholdings (Nakumatt
Holdings), which gives it excellent access to financing (investment loans), human
resources (rotating managers) and physical capital (such as trucks).

In 1995, inspired by first-hand experience with large, customer-oriented US retail
formats, Nakumatt introduced the supercentre store format into Kenya (i.e., department
stores with a fully-fledged supermarket added a la Wal-Mart). Today 8 of its 12 branches
can be categorized as supercentres. One of these, the Nakmnatt Mega branch in Nairobi,
has 175,000 sq.ft of floor space, making it the largest retail outlet in East Africa. The
large average size of its stores explains why, although it has less than half the number of
outlets of its main competitor Uchumi, Nakumatt has 30% more floor space (585,000 sq.
ft vs. 450,000 sq. fi). Throughout, Nakumatt’s consumer focus has been mainly on the
high-income segment (50% of its customers fall into this category). In 2001 Nakumatt
started selling F FV in its stores. Hesitant at first, it has now fully embraced the F FV
section and is rapidly expanding it; all its branches have a modern FFV section. Chain-
wide sales have grown to Ksh16m. per month and are expected to grow to Ksh50m. per

month within the next three years.

27

Second and Third Tiers

The second tier in the hierarchy consists of the medium-sized Tusker Mattresses, Ukwala
and Metro Cash ‘11 Carry chains with an estimated 6%,4% and 4% of the supermarket
sector respectively. All three are catering to and fiercely competing for the dollar vote of
the middle- to low-income urban consumers, Tusker and Ukwala by operating stores
located downtown near the busy bus stations used by consumers in this income class,
who do part of their shopping before going home at the end of the day, and Metro
indirectly by supplying the smaller self-service stores located near these same bus stages
as well as in the residential areas. The Tusker and Nakumatt chains are in a strategic
partnership which, given the two chains complementary customer bases, targets urban
households from the lowest to the highest income category.

The third tier, the remaining 38% of the supermarket sector, consists of a varied
group of smaller chains and independent stores. Here we find the supermarkets located in
the smaller towns as well as those that have traditionally catered to high-income groups
and expatriates (e.g., settlers’ stores). These stores compete with the above chains by (i)
moving to and expanding in understored areas, (ii) adjusting their product assortment
with Specialty foods for a particular target group (e.g., expatriates) and/or (iii) developing
a more personalized service. The small chains group consists overwhelmingly of 2-
branch and 3-branch chains (68% and 27% respectively). About 40% of these

supermarkets sell FFV.

28

3.4 Diffusion Patterns

Our research revealed several interesting patterns in the socio-economic and geographic
diffusion of supermarkets in Kenya. First, while Kenya’s white and Asian consumers
were important as part of the initial narrow base of consumers for the early stages of
supermarket development, we calculate that their purchases now constitute a mere 15%
of total supermarket sales.

Second, supermarkets have spread from higher-income niches in the capital, to
middle- and lower-middle income segments in the capital and out into secondary cities
and lately into towns. Table 3 shows that they are now found in urban centers of all sizes.
In 2003, nearly 60% of the stores were located outside of Nairobi and basically every
provincial capital had one or more supermarkets. Taking store size into account as a
proxy for sales, Nairobi still accounts for the majority (56%) of sales. In terms of store
density, it is clearly in the lead with 36 stores per million, nearly three times the level of

the rest of urban Kenya.

Table 3: Geographic Distribution of Supermarkets in Kenya, 2003

 

 

Urban centers Supermarkets Supermarket Density
floor space
Name or No. of Aggregated ‘000 (stores
pop. size centers population No. % sq. ft % per m.)
(‘000) (‘000)

Nairobi 1 2,600 94 42 1,190 56 36
Mombasa 1 800 10 4 220 10 13
100-350 18 3,320 66 29 480 22 20
50-100 30 2,670 33 25 190 9 12
25-50 37 1,730 17 8 50 2 10
10-25 33 760 5 2 20 1 7
All centers 120 1 1,880 225 100 2,150 100 19

 

Sources: population data extrapolated from 1999 census data (CBS, 2002b) using urban population
growth data fiom UN Population Division (2003) and UN Habitat (2004). Supermarket data based on
the author’s survey.

29

Figure 4 depicts the growth of supermarkets in terms of floor Space in Nairobi vs. other
towns over the period 1975-2003. In the absence of historical statistics on the number and
size of retail outlets, we used the year of establishment of the supermarket stores in
existence in 2003 as a proxy. The figure shows that supermarkets are growing almost

simultaneously in Nairobi and elsewhere.

Figure 5: Growth of Supermarkets in Nairobi vs. Other Urban Centers
in Kenya, 1976-2003

 

 

1,200,000

1,000,090 /

TotalSeleeFtoorSpece
tuft)
. . .§.§
. 9 s

 

_‘ ‘AAA-n

 

o ':°°.°'.'......-Te.......ffir.....-.
19701978198019821984199619881990199219941996199820002002
Year

 

 

 

[— Nairobi ------- Other Urban Centres .

 

 

Source: author’s national supermarkets survey.

Nevertheless, there appear to be two successive waves of diffusion. First, one observes
the emergence of independent stores which expand to other nearby centers, becoming
small local chains in the process. Then the top five chains move into town, creating a
shake-out whereby the out-competed smaller chains either close down completely or
move to the next, less competitive and (increasingly) smaller town, thus fuelling the first
wave of diffusion. From their base in Nairobi, the top five chains have moved into
Mombasa (population 800,000) and intermediate cities (population ZOO-350,000).
Uchumi moved to Nakuru in 1993, to Eldoret in 1999 and to Mombasa in 2000.

Nakumatt moved to Mombasa in 1996 and to Kisumu in 1997. Metro moved to Eldoret in

30

1999 and to Kisumu in 2001. Ukwala followed suit moving to Eldoret in 1999 and to
Kisumu in 2003. The top five chains are now also expanding into smaller towns: Uchumi,
as part of its five-year plan, opened three of its latest stores in smaller towns like Meru,
Karatina (both around 125,000 inhabitants) and Kisii (75,000 inhabitants) and announced
further plans for expansion into other small towns like Thika and Nyeri.

In their expansion from Nairobi, the supermarket chains select towns that have (i)
large populations (in the town centre and its hinterland), (ii) many households with a
regular income (implying more potential regular customers), (iii) nearby high-
consumption customers such as rich white farmers or major tourist resorts, and (iv) less
competition from other supermarkets. Table 3 shows that moving away from Nairobi to
intermediate and then smaller towns implies moving from high to lower supermarket
density (and thus less competitive towns). Growing out into underserved urban areas is
considered by the Kenyan supermarket sector a big opportunity — ‘a domestic
supermarket scene far from being saturated’ (Wachira, 2002; Wahome, 2001).

The growth of Kenya’s supermarkets has also taken on an East African regional
character with outward Foreign Direct Investment (FDI) facilitated by regional trade
agreements COMESA (Common Market for Eastern and Southern Africa) and (from July
2004) EAC (East African Community). Against the back-drop of a rapidly growing
regional trade (for example, since 1998 COMESA has outgrown the European Union as
Kenya’s most important export market), Uchumi opened its first branch outside Kenya in
Kampala (Uganda) in December 2002, a key motivation being the expectation of higher
margins there (Akumu, 2003). From their regional expansion plans for the next five years

it appears this is just the first drop of an upcoming flood of outward FDI by Kenya

31

supermarkets. When interviewed on the topic, Mr. Vijay Shah, Strategy Coordinator of
Nakrunatt, indicated that his company plans to open branches in Uganda (2), Tanzania
(2), Rwanda (1), Burundi (1), Zambia (1) and Zimbabwe (1) over the next five years.
Uchumi’s expansion plans include stores in Tanzania and Rwanda as well as more stores
in Uganda.

Such rapid diffusion of supermarkets — the start of a retail transformation in
Kenya — is bound to have, as it has had elsewhere, important effects on the agrifood
systems, and markets faced by farmers. To understand these changes we now focus on

changes in the procurement systems of the leading chains.

4. Supermarket FFV Procurement Systems and Requirements

Our conceptual approach to the changing structure of the supermarkets’ FF V
procurement system mainly draws from strategic management theory (SMT) and
transaction cost theory (TCT). Each theory addresses a different side of the same coin:
SMT deals with the objective of creating competitive advantage (benefit-Side), TCT
looks at optimizing the cost efficiency of the institutional format (governance structure)
used in creating competiting advantage (cost-side).

SMT postulates that economic agents develop and implement strategies which
reduce competition in the industry and thus enable them to earn above normal profits
(Porter 1980, Barney 1996). They do so by (continiously) creating competitive

advantages5 which competitors cannot easily imitate or nullify because of barriers to

 

5 Examples of such competitive advantages are: lower production costs, higher perceived product quality,
the development of products with unique benefits that are highly appreciated in the market.

32

entry (e.g., economies of scale, product differentiation, collusion in a concentrated
industry, and so on). The extent to which firms can develop such competitive advantages
critically hinges on how (well) they are linked to their input and output markets. In the
TCT framework (Williamson 1985, 1991), economic agents (in our case the supermarket
produce procurement managers) choose vertical coordination strategies that minimize the
transaction costs which follow from having to deal with uncertainties caused by a
combination of the agent’s bounded rationality, the (potential) opportunism of their
trading partners (in our case the FFV suppliers) and the presence of (physical and
organizational) assets which are highly specific to the transaction (i.e., these assets loose
much of their value in their next best alternative use). Vertical coordination strategies can
be set out along a continuum from spot market transactions (high in flexibility, low in
control) to vertical integration (high in control, but danger of x-ineffrciencies) (Peterson
et a1. 2001).

Before the rise of domestic supermarkets, there existed two distinct F F V
marketing systems, one at each of the extremes of the vertical coordination continuum:
the traditional domestic system and the export system. Kenya’s traditional FFV
marketing system is characterised by fragmentation at both the producer and the retailer
ends of the supply chain, market power with the wholesalers, long channels, direct
payment to suppliers, little quality control or grading, few standards, little or no product
innovation, small volume transactions and small inventories. In stark contrast, the export-
oriented part of the horticultural sub-sector is highly dynamic, continuously developing

new products and implementing institutional and organisational changes (such as

33

contracts, codes of good agricultural practices, traceability systems) that have increased
competitiveness and spurred a rapid growth of exports to the European Union market.

The introduction of a fresh F FV section by the leading supermarket chains in the
second half of the 19905 created a third marketing system for FFV in Kenya, which over
time is shifting from the traditional to the export model. At first, supermarkets relied on
traditional wholesale markets, basically taking the traditional informal sector supply
chains of FFV as givens and working with these. This is the way the non-leading chains
still work today, and we do not therefore deal with them in this section. Overall,
wholesale markets remain the main source for these supermarkets, with on average 29%
of supermarket FFV supplies coming from this source. Next are direct supplies from
farmers (23%), supplies from brokers (19%) and supplies from importers (11%). The
remaining 18% consist of supplies from own farms and from exporters.

The F FV section in Kenya’s supermarkets has played an important role in their
competitive strategies. At first, simply having a F FV section was the competitive
advantage, this then shifted to having a larger assortment, better quality, more reliable
availability and more value-added products (as the initial focus was on high-income
consumers), followed by the current focus on getting prices down (in order to reach
middle-to-low income consumers) and plans are being drawn to make food safety
guarantees the next competitive advantage in the FF V retailing by supermarkets. In
aiming for these different objectives, supermarkets face many uncertainties when
sourcing form Spot markets: quality, availability, price, and so on, are all fluctuating
strongly. In order to gain more control over their supply chains, supermarkets have

started shifting away from spot markets along the vertical coordination continuum. While

34

vertical integration (supermarkets owning their own farms) has emerged in Kenya’s
supermarket sector, it is rare and appears to be destined to disappear as those cases
indicate that the costs of organizing such disparate activities under one management are
high. The governance structures supermarkets are moving toward most clearly are
therefore contract relationships (intended to be long-term) which provide more control
than spot markets but keep overhead costs down and provide greater operational
flexibility. A similar observation on the importance of long-terrn contractual relationships
was made by J affee (1995) with regard to Kenya’s FFV export channel.

Both to create competitive advantage and to manage transaction costs,
supermarkets have implemented a number of technological, organizational and
institutional changes: formal quality standards (and their enforcement mechanisms),
contracts, distribution centers (centralized buying), IT systems for product flow
management and communication, Shorter supply channels (more direct links with
farmers) and fewer (but larger) suppliers (like their customers, supermarkets prefer one-
stop shopping for their procurement). As these chains grow, they are building up the
market power, financial means and geographic presence to realize all these changes and
to demand compliance from their suppliers. The supermarkets’ FFV suppliers on the
other hand have to make transaction specific investments of their own (e.g., organize
their labor force to deal with short order cycles, develop staggered production systems
which smoothen out supply over time, and so on) and are therefore also keen on
establishing contractual relationships which help safeguard these assets to some extent.

With supermarkets and their FFV suppliers continuously increasing the sophistication of

35

their supply chain in the pattern set out above, they are creating a fundamental structural
change in agri-food supply chains.

In the remainder of this section we focus mainly on the two market leaders
(Uchumi and Nakumatt, which sell 90% of FFV sold through supermarkets), in order to
illustrate first the difference and then the convergence of their procurement systems. A
few years ago, the two leading chains had very different systems (Uchumi with a
decentralized system with a preferred supplier program that emphasizes the use of
medium/large farmers, versus Nakumatt, with a centralized system but dependence on an
outside specialized wholesaler who relied on small growers). Driven by the same
determinants (create competitive advantage, lower transaction costs), the two systems
have very recently converged, however, with Nakumatt bringing procurement in-house
(into the same holding company) and shifting towards medium/large farmers to ensure
quality and consistency and reduce transaction costs. For supermarkets, smallholder FF V
producers Operating in groups could theoretically provide the same advantages as larger
farmers but few such groups exist for domestic market FFV, mainly because organizing
smallholder farmers in efficient and effective groups has substantial costs associated with
it and few intermediaries have (by 2003) taken initiatives in this regard. However, the
thousands of smallholder producers involved in Kenya’s (highly efficient) export
channels illustrate that it is possible (Jaffee 1995).

From an initial total reliance on traditional suppliers, the leading supermarket
chains have thus started to develop the five pillars that have characterized the evolving
FFV procurement systems of supermarkets in developing countries that are ahead of

Kenya in terms of supermarket development (Reardon et al., 2003): namely, (i)

36

centralization to distribution centers away from store-to-store deliveries; (ii) shift from
reliance on traditional wholesalers to use of Specialized/dedicated wholesalers; (iii) Shifi
from the spot market to use of preferred supplier systems; (iv) shift from local
procurement to regional, or regionalization; and (v) imposition of private safety and

quality standards.

4.1 Uchumi: A Decentralised System with a Preferred Supplier Programme6

C entralization
FFV procurement at Uchumi is in a transitional phase from semi-decentralised to
centralised. The retailer has most of its FFV suppliers deliver direct to its four
hyperrnarkets and to the City Square branch (which housed its headquarters until 2003) in
Nairobi. These five stores function as distribution centres to the smaller stores in Nairobi
as well as the up-country branches. This system applies mostly to high-volume, semi-
perishable items sourced domestically (e. g., onions, tomatoes, potatoes), imported fruits
(e.g., citrus from Israel, South Afiica, Egypt) and even some imported vegetables (e.g.,
garlic from China). At the same time, the smaller supermarkets in the chain have farmers
supply some FFV items (such as greens) directly. The use of dedicated F F V trucks (with
refiigeration) is not yet economical for the up-country branches, given the small volumes
involved.

During 2003, Uchumi constructed a new (large) distribution centre (DC) of

15,000m2, two units of which (one incoming, one outgoing) are equipped with cooling

 

‘ The information for Uchumi’s procurement system is mostly based on several interviews with Mr Nzioka
Kioko, Category Manager — Fresh Produce Uchumi, his successor Mr Peter Nderu and first-hand
observations by the author, interviews with branch managers and the author’s national survey.

37

facilities and intended for F F V. Improving quality level and consistency is the key reason
for centralizing the F FV procurement. In combination with a modern IT system that links
all of its stores by satellite communication (also installed in 2003) and that allows a
centralized tracking of individual product items (SKUS) at each store, the DC will allow
for better stock management of all SKUS (including the FFV which are sold bar-coded).
With the DC facilitating the development of cold chains, Uchumi plans to introduce
insulated trucks in 2004/5, at which point it will gradually start demanding investment in
cold chain technology from its suppliers. All FFV items are delivered by the different
suppliers directly and shelf-ready to the different procurement points. No packaging,

labeling or primary processing (e.g., fresh-cut) is done by Uchumi.

Supplier selection

As most Kenyan horticultural producers lack the scale of operations to provide the large
supermarket chains consistently throughout the year with the volurhe and quality of F FV
they require, Uchumi’s procurement managers purchase from a mixture of suppliers,
including farmers of different Sizes, traditional brokers and the wholesale market,
specialised wholesalers, importers and exporters diverting their (excess or quality
rejected) FF V to the domestic market. Table 4 indicates the composition of Uchumi’s
FFV suppliers in 1997 (the year of the first sales of FFV at Uchumi), in 2003 and in 2008
(as estimated/predicted by Uchumi’s produce manager). In this table, ‘farms’ refers to
farnrs that supply Uchumi either directly or via a Single marketing facilitator who has a
longer-term relationship with them. Examples of such market facilitators are farmer

marketing associations and specialised wholesalers who work with farms through

38

outgrower schemes (e. g., exporters, lead farmers). Marketing facilitators are still rare
amongst Uchumi’s FFV suppliers (some examples are given below) and in this
description we group them under ‘direct supplies from farmers’. By contrast, traditional
brokers and wholesalers are those who bulk supplies from many sources without
registering or preserving the identity of the farms. We define ‘small farms’ as farms with
less than 10 acres, ‘medium-sized farms’ as farms with 10 to 40 acres, ‘large farms’ as
farms with 40 to 200 acres and ‘plantations’ as farms with more than 200 acres.
According to Table 4, for vegetables, which make up 45% of the value sold at
Uchumi, roughly 50% is sourced directly from growers. Medium-sized producers supply
the largest share, with 25%, followed by large farms with 15%, and small farms with
10%. Brokers supply 45% of Uchumi’s vegetables, while the rest (5%) is imported. Small
farmers supply mostly leafy greens (kale, spinach, traditional African vegetables) and
vegetables sold in small volumes (e.g., herbs). Other vegetables are supplied by the larger
farmers. The latter especially applies to fresh-cut vegetable packs because most small-
scale farmers do not have a packing shed, which in this case is a key requirement.
Currently 75% of fresh-cut vegetable packs are supplied by large farms and this
percentage is expected to increase to 90% over the next 5 years. Brokers mainly resolve
shortfalls. Imported vegetables include tomatoes, onions, garlic — basic products for

which local supply chains are still inadequate relative to Uchunri’s needs.

39

Table 4: Uchumi's FFV Supply Composition by Supplier Type, 1997-2008 (%)

 

 

. Vegetables Fruits
Type °f “1’9"” 1997 2003 2008 1997 2003 2008
Small farms 13 10 15 5 10 10
Medium-sized farms 10 25 30 10 10 10
Large farms and plantations 5 15 35 0 15 35
Traditional brokers/wholesalers 70 45 10 70 40 10
Imports 2 5 10 15 25 35

 

Source: author’s interviews with Mr. Nzioka Kioko, F FV Manager, Uchumi.

For fruits, which make up 55% of the value sold, Uchumi sources 35% directly from
growers, 15% from large-scale farms, 10% from medium-sized farmers and 10% from
small producers. Imports represent roughly 25% of procured fruit and the remaining 40%
is supplied by brokers. Small farms play only a small role with regard to fruits (examples
of fruits where they are involved are watennelons, passion fruit and strawberries). For
fruits there is a heavy reliance on brokers (because they buy mangoes, for example, from
smallholder producers in different regions of the country as the seasons change), large-
scale farms/plantations (e.g., Kakuzi, a 6,400-acre agrifood business listed on the Nairobi
Stock Exchange) and imports. As a group these three suppliers represent 80% of
Uchumi’s fresh fruits supplies.

As Uchumi’s sales of FFV increase, it is moving away from traditional brokers
(and their long supply chains and the mostly smallholder producers they buy from) to get
supplies directly from farmers. Table 4 indicates that brokers, as a source of F FV, have
decreased from the main supplier category (70% in 1997) to less than 50% in 2003 (45%
of vegetables and 40% of fi'uits). Reducing its reliance on brokers is the first priority at
the moment for Uchumi‘s F FV procurement, and management expects that by 2008
brokers will make up no more than 10% of supplies, i.e. they will only be used to resolve

shortfalls from regular suppliers (similar to F reshmark, Shoprite’s F FV procurement arm

40

in South Africa; Weatherspoon and Reardon, 2003). Direct supplies by farmers allow
supermarkets to increase simultaneously control over quality, supply reliability and price
stability and thus make them more competitive with traditional retailers (in terms of
product quality and reduced stock-outs).

Given the increase in scale and the need to be price-competitive, the supermarket
FF V procurement managers are under cost pressures to deal with fewer and larger
suppliers. Table 4 indicates that large farms are the main beneficiaries of Uchumi’s
declining reliance on traditional brokers. Between 1997 and 2008 large farmers are
expected to increase their share in Uchumi’s vegetable supplies 7-fold from 5% to 35%.
Absent as suppliers of fruits in 1997, large farms now supply 15% and are expected to
have a 35% share by 2008. A notable example is the 13,000-acre pineapple plantation
that is a vertically integrated part of Delmonte, a pineapple and other fruits processor
(canned fruits, tetrapak-packaged fi'uit juices). The key attraction for a supermarket chain
like? Uchumi is that a company like Delmonte is already used to dealing in large volumes,
assuring a consistent volume of high quality supplies for its processing plant.

Uchumi already relies heavily on a small group of about a 100 medium to large
suppliers, especially when supplies are low. These 100 farmers form the core of an
emerging preferred suppliers list. All in all, there were about 150-220 regular, registered
suppliers for FFV at Uchumi in 2003 (depending on the season). This number is expected
to rise to 300 over the next four years, mainly because the planned network extension to
50 stores will include several up-country branches that will add suppliers from the local
farming communities. Given the current decentralized procurement system, branch

managers can (but are not restricted to) use this list of registered suppliers to order F FV

41

in a composition that fits their clientele (payments to suppliers are still dealt with
centrally). As a group, medium and large farms already make up 40% of vegetable
supplies and 25% of fruit supplies. By 2008 they are expected to become the dominant
suppliers with 45% of fruit supplies, and 65% of vegetable supplies. Larger suppliers are
not only gaining ground because they can supply larger volumes and thus reduce
transaction costs for Uchumi. These farms are also the ones which have invested in
irrigation systems which give them greater control over quality as well as allowing them
to produce all-year-round. One reason why Uchumi prefers direct supplies by farms is
that it allows them to inspect them and observe their irrigation system first-hand. Having
reliable irrigation is a sine qua non for farmers who want to get on the supermarkets’
preferred supplier list.

Marketing facilitators or wholesalers who are Specialized (in a particular product)
and/or dedicated (to a particular supermarket), are rare amongst Uchumi’s suppliers, but
they are emerging. Currently they focus on smaller-volume F FV items that address
particular gaps in Uchumi’s F F V product line. The following are some examples: Family
Concern, Iga Muka, and Sunripe. Family Concern is an NGO that aims to combine
development and business objectives in building horticultural linkages. Having identified,
besides Uchumi, that there exists a high, unmet demand for traditional African
vegetables, Family Concern set out to build supply capacity for traditional African
vegetables amongst smallholders. These products appeared in the F FV section of
Uchumi’s Ngong hypermarket in 2004. Iga Muka, a 30-member farmers’ association
operating in the Sagana development scheme on the slopes of Mount Kenya, succeeded

in linking up with Uchumi via the Kenya Agricultural Commodity Exchange (KACE), a

42

private sector firm that facilitates linkages between sellers and buyers of agricultural
commodities, mainly by developing marketing information systems that reduce the
information asymmetry between farmer and broker. Iga Muka supplies small volumes
(60-100kg/week) of strawberries to Uchunri. Sunripe Ltd, one of Kenya’s leading agro-
exporters, produces from its own farms, but also buys from around 1,000 smallholder
producers, predominantly as outgrowers through contract farming (Shah, 2004). Sunripe
is Uchumi’s key supplier for pre-packs of mixed vegetables. It is used to the high
demands (for example, with regard to hygienic processing) and contracts often for more
volume than it can sell in export markets. Since it is contracted to buy the FF V from the
outgrowers, it needs a market for the excess production. Uchumi gets a high quality
product while Sunripe gets more flexibility, while spreading its market risk.

Uchumi’s historic relationship with the government (formerly govemment-owned
and still with a strong government presence on its board) has had little impact on its
selection of F F V suppliers, not only because Uchumi operates like a regular commercial
player which has to respond to the challenges of a competitive market, but also because
the government has largely been absent from the horticultural sub-sector.

Uchumi’s selection of F FV suppliers is also determined by their business
management skills. Compared with selling to traditional markets, selling to supermarkets
implies far more formal transaction methods as well as more stringent delivery
conditions. However, most relationships are informal — only a small number (about 5%)
of suppliers have written contracts with the chain. These contracts stipulate the delivery
volume and frequency (similar to a standing order), the quality standard for the product

and (a limited set of) code of practice requirements such as non-use of banned chemicals,

43

the use of dust covers during transportation and use of clean irrigation water (as opposed
to sewage water). Although a running average price for the year is indicated, the price is
not fixed in the contract as Uchumi must be able to follow price movements in the
market. Because Uchumi wants to build up long-term relationships with its farmers, its
pricing policy is never to pay below the actual production cost (even if market prices
drop below cost) and to balance this by paying lower prices when prices peak. Contracts
also stipulate what constitutes an accepted reason for non-delivery. At this early stage in
the formation of the supermarket’s produce procurement system, these contracts are used
more as an instrument to streamline and formalize the relationship between supermarkets
and their preferred suppliers, than that it is intended as an enforceable legal document. At
this stage both the farmer (who faces the risk of not being able to supply the required
volume) and the supermarket (who risks having to pay farmers a price that is too high
above fluctuating wholesale market prices to be competitive with traditional retail
outlets) prefer contracts (in terms of volumes and prices) that have a great deal of
flexibility.

Our interviews with management revealed that the percentage of farmers under
contract is planned to increase dramatically to 70% within the next five years as key
suppliers (who have proven their value) are identified. These key suppliers are expected
to be distributed as follows: 30% as medium-sized farms, 40% as large farms and 30% as
importers. Contracts will also focus mainly on products for which there exists a
resolvable supply problem. For example, contracts are not needed for local pears, a
seasonal item which is available in great quantity during a brief period of a few months

and totally unavailable during the rest of the year.

44

Uchumi reduces transaction costs by requiring that listed suppliers have cell
phones. About 85% of orders are placed by telephone — even to smaller producers — and
for the most perishable F FV (e.g., the leafy greens) these orders are placed only 12 hours
before expected delivery (evening call for next morning delivery). Suppliers should also
have bank accounts, as Uchumi plans to move to payments through bank transfers on a
monthly or bi-weekly basis. Suppliers without phones and bank accounts will be delisted
in future.

Uchumi pays 2-4 weeks after delivery rather than the cash-on-delivery payment
of the traditional systems. Suppliers to Uchumi also pay fees for breakages (unsold or
Spoiled produce) and for promotions. Before the introduction of these breakage fees (a
flat rate of 7% of the purchase price), Uchumi used to inform farmers of the amount of
breakage and ask for replacement at the farmer’s cost — a system which was abolished
because it increased transaction costs and suffered from the practical problem that it was
difficult to know which supplier’s produce was responsible for what percentage of the
breakage. In terms of delivery conditions, supermarkets are much more stringent than the
traditional system. For example, FF V needs to be washed and sorted (some sorting takes
place at the receiving bay, but for the greater part if the product is not graded, the whole
load is rejected). It also has to be supplied regularly (e.g., daily for tomatoes and greens,
every 2-3 days for semi-perishables like onions), has to weigh an assured minimum,
needs to be transported with great care (e.g., leafy vegetable bundles need to be
transported loose, rather than stuffed into bags as is required traditionally) and, if
packaged, needs to be bar-coded and labeled. Although not required at the moment,

packing sheds will become compulsory.

45

It is evident that several benefits outweigh the above non-traditional costs and
requirements of supplying to Uchumi, or this market would not attract medium-sized and
large farmers that have market options. Namely, Uchumi lowers access costs and
requirements for a variety of inputs. (i) Uchumi intermediates with seed, fertilizer, and
chemical suppliers to negotiate lower prices and better quality for its preferred suppliers.
(ii) Uchumi also intermediates with creditors for loans for investment in irrigation and
trucks — and thus a written contract with Uchtuni serves as a kind of collateral substitute.
These loans can be substantial - from Ksh400,000 to Ksh6m7. In this context, Uchumi
has been negotiating with organizations like the International Finance Corporation in
order to provide better lines of credit to its FF V suppliers. Once supplier-contracts
become more widespread, FFV suppliers will find it easier to access credit through
regular charmels as it will be easier to convince loan administrators at commercial banks
of their creditworthiness. (iii) Uchumi reduces information costs by organizing meetings
with its FFV suppliers to inform them of market opportunities and requirements; it is also
considering providing direct technical assistance to its preferred suppliers (via a full-time

agronomist).

Regionalization

Although horticulture is well-developed in Kenya, especially for vegetables, and most
FFV items are procured locally, imported F F V are gaining importance on the
supermarket shelves. As Table 4 shows, imports are expected to increase their share of

Uchumi’s vegetable supplies S-fold fi'om 2% in 1997 to 10% in 2008. For fruits, imports

 

7 For comparison, the average value of land in our sample was around Ksh600,000 per acre, with
smallholder farms having an average size of 5 acres. These loan amounts therefore are of a similar
magnitude as the value of the land owned by smallholder farmers (usually their only possible collateral).

46

are even more important, with 25% of supplies today originating from other countries.
For F FV as a category, imports make up about 15% at Uchumi, at least three times the
share of imports in the country’s urban FF V market as a whole (estimated at less than 5%
of total consumption).

Imported FFV are becoming more important for several reasons, depending on
product and season. (i) They complete the year-round availability of items which are not
(e.g., apples), or only during a few months in the year (e.g., pears), produced in Kenya.
(ii) They are more competitively priced and address low availability due to seasonality
(e.g., the mostly informal imports of pineapples and bananas from Uganda and tomatoes,
oranges and onions from Tanzania). (iii) They offer produce of a quality which is not
available from producers in Kenya (e.g., oranges from Israel, bell peppers from Holland).
In some cases, all of these reasons apply simultaneously. For example, garlic from China
is cheaper, available year-round and of a superior and more consistent quality (at least
visually) than the locally grown alternative. As a consequence, imported garlic, available
in Kenya’s supermarkets and open air markets alike, has grown from 0 in 1997 to 645MT
in 2002 (FAO, 2004).

These reasons are particularly important to supermarkets that derive their
competitive advantage over traditional retailers from their broad assortment, better
quality, year-round availability and (for FFV to a lesser extent) lower retail prices.
Furthermore, Specialized importers provide a structure supermarkets are more familiar
with from their dry foods retailing: one-stop order-points which can offer regular supplies

throughout the year and are accustomed to formal business methods. Uchumi does not

47

import FFV directly, but relies on importers, as it does not yet deal in large enough
volumes of these items.

Supermarkets, as they grow regionally, will look for suppliers from an
increasingly wide pool, becoming regional traders in the process. For example, Uchumi’s
hypermarket in Kampala is closely linked in a bi-country procurement system with the
Uchumi chain in Kenya.8 Various FFV items are exported from Kenya to the Kampala
store, most notably vegetables which are not as readily available in Uganda as they are in
Kenya (such as carrots, cauliflower, herbs, pre-cut Asian vegetables). About 40% of the
FFV sold in the Kampala store is imported from Kenya or elsewhere. At the same time,
Uchumi is looking for opportunities to export F F V items from the 60% they source
locally back to Kenya. This is considered desirable by the Ugandan government, in order
to prevent the arrival of a Kenyan chain further exacerbating trade imbalances and the
flooding of the Ugandan market with agri-food products from Kenya, but rather
representing a ‘win-win’ situation where Uchumi helps to build a procurement system
that relies on growers in both countries. No Uganda-to-Kenya trade exists within
Uchumi’s FFV procurement system at this time, although pineapples imported through
traditional trading channels/brokers end up on its shelves in Kenya. The extent to which
this regional sourcing will become important depends on many factors, most notably (1)
how fast the chain will regionalize, (ii) how fast it will modernize its FFV procurement
system (especially with respect to centralization and cold chains) and (iii) the nature of

suppliers found in regional markets outside Kenya.

 

' The information in this paragraph is based on personal communication with Mr Bernard Kibaru, Uganda
Country Manager of Uchumi, Kampala, June 2003.

48

Safety and Quality Standards
Since it first started selling F FV, Uchumi has required its preferred suppliers to dedicate
all their top-grade output to it (they can sell lower grades to the traditional channels).
FFV quality standards are ‘private’ (specified by the buyer, not the government) and are
for the greater part limited to appearance (colour, size, blemishes, etc.). Produce is
inspected visually by Uchumi personnel at the point of delivery (the five procurement
points above). For example, for tomatoes the (current) Uchumi quality standard consists
of about 20 specifications including for cleanliness, size, shape, weight, variety, and so
on.9 Produce not meeting the quality standards is rejected. About 10% of FFV is rejected,
due to failure to meet quality requirements. Of this rejected F F V 45% comes from small
farms, 45% from brokers and 10% from medium-sized farms and importers (produce
from large farms is rarely rejected). Over time suppliers learn which quality is acceptable
(it varies with seasonal availability; i.e., standards are lowered when supplies are low)
and they work towards supplying this quality only to supermarkets.lo Unlike the
traditional marketing channels, selling a lower quality for a lower price is not an option.
Quality standards are applied at reception from preferred suppliers without
contracts, or imposed within the contracts for the small (but rapidly increasing, as noted
above) share of suppliers that have contracts. For the latter, Uchumi uses the measures for
reducing quality risk identified by Hueth et al. (1999) in the case of FFV contracts in

California: (i) monitoring farms in the field (e.g., irrigation system) to see if they have the

 

’ Similar standards are already developed (for some products, e. g., for tomatoes since 1985) by the Kenya
Bureau of Standards, but no case of their implementation in the domestic market was encountered or
mentioned in the course of our research.

'° This point was confrrmed in our interviews with Uchumi suppliers, several of whom did mention,
however, that standards were at times misused by individual receiving agents to reject their produce in
favour of produce supplied by a broker with whom the agent has an informal business relationship.

49

capacity to produce FFV of quality; (ii) promoting management practices that lead to
quality by specifying which variety farmers Should grow, even assisting them to get the
right seed quality to grow; (iii) measuring quality directly at the point of delivery; and
(iv) sharing price risks with farmers (indirectly) via breakage fees.

Neither Uchumi nor any other supermarket in Kenya has developed its own
private food safety standards; nor does Uchumi implement public food safety standards
for F FV. Public health and food safety standards exist; there is in fact a Kenya standard
Code of Practice for the horticultural industry (derived from the export industry) which
deals with hygiene, worker welfare, environmental protection, and so on. However, our
interviews with various economic actors indicated that these food safety/process
standards are not (or minimally) implemented in practice (limited to, for example,
cleanliness inspections of city markets). Nevertheless, Uchumi inspects the farms of key
suppliers, inter alia to make sure no sewage water is used for irrigation (a practice
observed in urban agriculture). This is a guarantee the brokers/wholesale markets cannot
provide to their customers.

Another example of how food safety issues begin to come into play is that
Uchumi is hesitant to procure from small farmers because they are unlikely to be able to
invest in packing sheds (required for hygienic handling of produce). These sheds are not
yet required by Uchumi, but before developing long-terrn relationships Uchumi wants to
make sure that the farm has the capacity to grow with them. A third example illustrating
the emergence of food safety monitoring at Uchumi is the specification for tomatoes
supplied to Uchumi that they should be ‘free of chemical residues’. However, detailed

food safety process standards for FFV (such as the selection and use of pesticides by

50

farmers) are not yet employed by Uchumi or any other supermarkets in Kenya, nor are
products sample-wise tested for bacterial or pesticide residues. However, Uchumi
believes that if it could provide a safety assurance to its customers, it could triple its
current FFV sales. These FFV items could likewise be sold under the Uchumi label
(already used for the Asian pre-cut vegetable packs supplied by companies like Sunripe
which comply with strict EU standards). F FV suppliers who meet high quality and safety
standards (similar to those used in the export sector), and who can produce competitively

for the domestic market, have substantial market opportunities.

4.2 Nakumatt: A Centralised Procurement System Relying on Procurement
Outsourcing through Specialised/Dedicated Wholesalersll

Like Uchumi, Nakumatt built (and started using) a new DC in 2003. Unlike Uchumi,
however, the retailer has no plans for moving produce through it, as it continues to rely
on specialised wholesalers, who de facto represent FFV distribution centres. This use of
specialised wholesalers is driven by two factors. First, Nakumatt started selling FFV with
some reservations at first (in 2001, four years after Uchumi) when it had only seven
branches, which implied that volumes were too small to warrant an in-house procurement
system. Second, Nakumatt employs a strategy of outsourcing certain product lines (e.g.,
clothing, furniture, books, bakery and FFV; about 30% of its sales), while it focuses on
its core competence, namely, building up a strong retail brand. Firms to which such

activities are outsourced operate on a consignment basis: (i) they manage the store space

 

6 The information for Nakumatt’s procurement system is mostly based on interviews with Mr Abdul Sidi
(Managing Director) and Mr Rayhon Aswani (Assistant Manager) of Mugoya Vegetable Shop, Mr
Mahesh (Manager) of Fresh ‘11 Juici, and first-hand observations by the authors, interviews with branch
managers and the authors’ national survey.

51

Nakumatt is willing to dedicate to the product category (e.g., the F FV section), retaining
ownership of the stock; (ii) their product is sold by Nakumatt, which acts as their
mercantile agent; and (iii) they are paid the sales revenue minus a commission (an agreed
percentage of sales) by Nakumatt. In the case of FFV, this commission is currently 20%
of sales and covers the use of the Space (rent, water, electricity, etc.), access to the
customer base, sales activities and a profit margin for Nakumatt supermarkets.

Nakumatt’s specialized wholesalers are Mugoya Vegetable Shop, in November
2003 replaced by Fresh ‘11 Juici Ltd, for the Nairobi and Kisumu branches and Shree
Ganesh for the two Mombasa branches. The need for two wholesalers is a direct result of
the absence of cold chains in the domestic marketing of FFV. While transportation at
night through the higher (and cooler) sections of western Kenya towards Kisumu is
feasible in terms of preserving produce freshness, this is less the case when transporting
to the more distant, hotter and more humid Mombasa region.

Mugoya and Shree Ganesh are greengrocers which over the years have built up a
portfolio of institutional customers (schools, hotels, restaurants, government
organizations, and so on) whom they supply directly. Mugoya, for example, has about 80
of such institutional customers. Catering to such a heterogeneous group, who need daily
supplies in small volumes of a wide variety of fruits and vegetables, had given these
wholesalers the capacities Nakumatt was looking for in 2001 when it wanted to sell FFV
in its stores. However, by 2003, Mugoya and Nakumatt were no longer satisfied with
their relationship. Mugoya felt that too much of its resources was going into supplying
Nakumatt (relative to the revenues) at the expense of its relationships with its institutional

customers. (Between 2001 and 2003, Nakumatt’s share in Mugoya’s sales grew from 0%

52

to 40%, with most of the remainder going to institutional customers; a small percentage
was sold through its own shop and to Tusker, Nakumatt’s strategic partner.) Nakumatt,
on the other hand, wanted a faster growth of the FFV section as well as a higher
commission. In November 2003, both parties separated amicably and Mugoya was
replaced by Fresh ‘11 Juici Ltd, a wholly-owned subsidiary of Nakumatt Holdings, set up
in 2003 as a dedicated supplier of fruits and vegetables to Nakumatt supermarkets (this is
similar to F reshmark for Shoprite in South Africa, see Weatherspoon and Reardon,
2003). Like the latter, it benefits from its embeddedness in the Nakumatt network which
offers enhanced access to financial, human and physical capital. The remainder of this
analysis of Nakumatt’s F FV procurement system will focus on its two (successive)
Nairobi-based wholesale suppliers, Mugoya and Fresh ‘n Juici, allowing us to track its
evolution over time.

Mugoya’s procurement system differs from Uchumi’s in two main areas: (1) it has
a stronger reliance on supplies from smallholder producers; and (ii) it has a broad-ranging
set of in-house value-adding activities (from sorting to fresh-cut packs). Table 5 provides
some detail on their supplier composition.

As Table 5 indicates, Mugoya, which has about 400 suppliers, purchases 60% of
its fruits and 70% of its vegetables from small producers. Although most of these
suppliers are located near Nairobi, Mugoya tries to reduce its supply risk by having
suppliers from different regions. This heavy reliance on smallholders is a part of a
broader structure which, aside from the fact that farmers supply directly to Mugoya’s

central facility in Nairobi, is not very different from the traditional system.

53

Table 5: Nakumatt/Mugoya's FFV Supply Composition by Supplier Type, 2003 (%)

 

 

Type of Supplier Vegetables Fruits
Small farms 70 60
Medium-sized farms 10 5
Large farms and plantations S . 5
Traditional brokers/wholesalers 14 20
Imports 1 10

 

Source: author’s interviews with Mr. Sidi and Mr. Aswani, Mugoya Vegetable shop.

Relationships between Mugoya and its suppliers are not generally geared to the long
term. Some farmers supply irregularly (in some cases only once or twice in a season) and
just pass by Mugoya’s facility at that time. Others have ongoing orders (e.g., 10 tons of
cabbage a week) for which prices and volumes are adjusted on a weekly basis. Changes
in orders are given during a previous delivery or are called-out. These orders are mostly
given to a group of 35-40 larger producers (i.e., producers with 5-10 acres or more) or
small-scale farmers who act as part-time brokers and bulk produce from neighboring
farms. For example, there are 3-4 small-scale suppliers for cabbages, with some dropping
out from time to time. No written contracts are used, partly because, according to
Mugoya, they are too risky since farmers cannot guarantee volumes or do not stick to
contracts, while Mugoya cannot guarantee prices. Within this context, these small-scale
farmers, while mostly regular suppliers known to Mugoya, are not listed in a preferred
suppliers list. Traceability is not an issue at this time.

Quality standards (as in written product Specifications) are not used and products
are not necessarily washed or sorted before delivery. Mugoya buys (in principle)
everything farmers supply, as long as a price can be agreed. It can afford to buy FF V of a
range of quality levels because it does its own sorting and has a range of customers with

varying quality preferences. The best quality went to Nakurnatt’s FFV section because

54

this is where (mostly high-income) customers base their buying decision in part on how
fresh and good the product looks. Poorer quality goes into its own processing and to the
kitchens of its institutional buyers (who process part of the FFV). Mugoya is also
responsible for taking back produce which is no longer fresh enough for display in
Nakumatt’s FFV section (in the consignment system it retains ownership of the produce
until it is bought by a consumer) but does not charge the cost to the farmers. Quality
permitting, some of the produce taken back can also be supplied to institutional buyers at
reduced prices (in order to avoid a total 1055). Although the firm has cold storage, this is
used to store imported fruits, overnight storage of left-over produce and internally
processed FFV, but not as part of a cold-chain structure. Farmers are paid cash-on-
delivery at prices which follow wholesale markets very closely. While this system is less
demanding for suppliers than Uchumi’s and hence allows a greater participation by
smallholders, it also offers fewer opportunities for growth and therefore fewer dynamics.
Medium-sized and large farmers (15% of vegetables and 10% of fruits) are
important as they provide a dependable regularity in the supply of key items (e.g., kale,
pineapples). These are also the farmers who are listed more formally as suppliers.
Brokers (14% of vegetables and 20% of fruits) are only used in the case of shortages
from regular suppliers (mango supplies, for example, depend on brokers during a few
months in the year when supplies are low). Imports (1% of vegetables and 10% of fi'uits)
are less important than for Uchumi, but are of a similar composition (pineapples, bananas
and ginger from Uganda, apples, pears, plums, peaches, grapes, oranges, bell peppers
from South Afiica, Egypt, Israel or Holland, garlic from China, onions, oranges and

tomatoes from Tanzania).

55

Mugoya, which has 150 employees, does all its value-adding in-house in its own
centralized facility. It washes, cleans, grades and sorts the produce, ripens bananas, and
pre-packs fruits (e. g., apples, oranges) and vegetables (e.g., onions, potatoes). It also does
some primary processing, for example making pre-cut vegetable packs for supermarket
shelves (e.g., stir-fry, snow peas, French beans, baby corn), potato chips/fries for the
institutional customers, and fruit juices. In order to meet customer expectations with
regard to hygiene during processing, Mugoya is in the process of obtaining HACCP
certification (postponed momentarily because it contemplates moving to a new, larger
facility). This processing capacity implies that, from its single facility, Mugoya can
supply a supermarket like Nakumatt with all the product variety it needs to stock a fully-
fledged FFV section.

When Fresh ‘11 Juici took over from Mugoya in 2003, its procurement system
started off from a similar structure, and about half of Mugoya’s suppliers shifted to it (at
least for part of their supplies). It also produces all the processed vegetable packs it sells
in Nakumatt supermarkets in-house. Like Mugoya, no contracts are used and quality
compliance is limited to a visual inspection of the produce at the point of delivery. All
supplies are delivered at Fresh ‘11 Juici’s central facility in Nairobi, except for the (more
perishable) leafy green vegetables which are delivered direct by farmers to the stores. The
company has a cold storage facility and uses it in the same fashion as Mugoya.

However, Fresh ‘n Juici has made some strategic changes. First, while it did get
most of its current 150 suppliers from Mugoya, it did not inherit the latter’s large
smallholder supplier base. From our interview with Mr. Mahesh, Manager of Fresh ‘n

Juici, we learned that smallholders have all but disappeared as fruit suppliers and make

56

up only about 20% of vegetable supplies, the main reason being, according to Mr.
Mahesh, that they cannot manage the quality and volume requirements demanded. For
kale, for example, Fresh ‘11 Juici wants suppliers who can supply at least 300kg twice a
week. Examples of FFV items where smallholders are still involved are yellow passion
fruits, greens, traditional African vegetables and herbs. Smallholder producers are
replaced by large farms and imports with respect to fruits and by medium-sized farmers
for vegetables. Both in importance and function, brokers remain in a similar role to that
they had at Mugoya, i.e., they represent about 10% of supplies and are mainly used to
deal with shortages. In terms of supplier composition, the shift from Mugoya to Fresh ‘n
Juici implies a convergence of the development paths of Uchumi’s and Nakumatt’s FFV
procurement systems, at least in terms of the relative importance in the supplied FFV
volume of the different types of suppliers .

Second, Fresh ‘n Juici also shifted to a low margin-high turnover strategy.
Suppliers, still paid cash-on-delivery, received higher prices than they got under Mugoya
(i.e. close to the prices in the traditional market), while consumer prices in Nakumatt’s
stores were reduced. Our price survey, for example, indicated that where as Nakumatt’s
prices for a selection of FF V items/stores were on average 17% higher than Uchumi’s in
November 2003, they were only 7% higher in April 2004. Taking a new price promotion
into account, whereby consumers using Nakumatt’s SmartCard (a customer loyalty card)
get an additional 5% discount on F FV, the price difference almost disappears.

Third, its nature as a dedicated supplier to Nakumatt also facilitated a heavier
involvement of management on the sales floor. In combination with a substantial

investment in extensive personnel training (Fresh ‘11 Juici has 120 employees), this

57

increased attention led to a more attractive presentation of the F F V section. While it is
mainly a dedicated Nakumatt (and Tusker) supplier, Fresh ‘n Juici may well expand over
time to include institutional buyers as well (thus also reducing its losses due to
breakages). It has been given the objective (by the Nakumatt Holding group) of raising
FFV sales at Nakumatt from the Ksh8m. per month under Mugoya in 2003 to Ksh50m.
per month within 3 years. By April 2003 (only five months after it took over), Fresh ‘n
Juici was already well under way to reaching its objective as sales had doubled to

Ksh16m. per month.

5. Implications for Development Policies and Programs

5.1 Importance as a Demand-Side Base for Agricultural Diversification

Supermarkets were a tiny niche in Kenya only a decade ago, but due to their 18% a year
growth rate since 1995 continuously removing market Share from traditional retailers
along the way, they now have 20% of urban food retail. In a few years the urban
population share will pass 50% (it stands at 40% today, from 15% in the 19705), and at
current growth rates supermarkets will be among the key gatekeepers to Kenyan dynamic
urban markets in 5-10 years. This makes them increasingly important subjects of
development policy attention for government and donors focused on helping Kenyan
smallholders access dynamic urban markets and diversify their agriculture into higher-

valued products.

58

Driven by demand-side reasons (consumers are shifting from traditional retailers
to supermarkets), it is likely that the supermarket sector will continue to grow in Kenya
and to displace traditional retailers — that is, following the international pattern. AS the
chains grow, they are more able to implement the kinds of procurement system changes
discussed in this essay, which in turn drives down their costs so that they can charge
lower prices and undermine traditional competitors, and so further expand — a virtuous
cycle for them. There is already evidence that this pays off for them and for the urban
consumers; Essay 2 shows that supermarkets charge on average 5% lower prices
(compared with small shops) for processed foods, and although their prices are higher for
most fruits and vegetables, for key ‘poor consumers' foods’ such as spinach, the largest
chains already have prices at or lower than open markets for products of similar quality.

We posit that the above-described process of supermarket development is the start
of an ascending competitiveness vis-a-vis traditional retailers of fresh foods, as has been
seen in other developing countries for the same reasons. While Kenyan supermarkets
have moved quickly into staples and processed food markets (such as for flour and edible
oils and snacks), they have only begun to make inroads into the FFV market. This
follows an international pattern. FF V are an important market for the promotion of
agricultural diversification for small farmers, as the production of FF V is relatively free
of economies of scale, at least with respect to land size (although the production and post-
harvest processes are often demanding of various forms of capital, including human,
organizational, and physical capital). Despite Kenyan supermarkets having only just
started merchandising FF V, the volumes they handle, and the amount they procure from

Kenyan farmers, are rapidly approaching the importance of the export market (handling

59

35,000 tons versus 67,000 tons exported). Moreover, the lead chains are increasing their

FFV sales quickly, and working costs out of and quality into their procurement systems.

5.2 Challenges to Supermarket Procurement Systems in Kenya

The front-runner chains (in a concentrated supermarket sector) are moving towards a
procurement system converging to the system developing elsewhere in developing
regions (Reardon et al., 2003), including centralisation of procurement and Shifts from
traditional brokers to Specialised/dedicated wholesalers, and from spot markets to use of
preferred supplier systems, and to use of private quality standards with (barely) incipient
use of private safety standards. These shifis have meant, already in this early stage of
supermarket development in Kenya, that supplies come mainly directly from large and
medium-sized farmers and somewhat from small farmers, and to a moderate but
decreasing extent from brokers who, in turn, are supplied mainly by small farmers. These
patterns and trends are driven by competition and a need for chains to reduce costs and
enhance quality and consistency.

Over the next half decade, issues will probably arise with respect to supermarkets
continuing to source mostly from large and medium producers producing for the
domestic and export markets. As supermarket procurement volumes grow substantially,
that system will be strained, and there will be a natural need to extend the grower base
locally — or to import more.

On the one hand, there may be few obstacles to sourcing much more from Kenyan

large/medium farmers in the short to medium term. These farmers are indeed switching

60

from traditional to supermarket channels, but they are also increasing their overall
production, including production for supermarkets — so one could argue that they will be
able to handle increased sales to supermarkets in the future (essay 3). Moreover, in other
countries, such as Costa Rica, China, Chile and Mexico, we have seen medium/large
farmers who were focused solely on exports turn in recent years to being major, and
increasing, suppliers to local supermarkets as the profitability of such sales, and the
capacity of the supermarket procurement system, have grown. Some of this growth in
Kenya will, of course, come through commercial outgrower schemes which in general
engage small growers. In fact, in places like Mexico much of the small farmers’
involvement in the supermarket channels has been via grower/shippers who have their
own production as well as outgrower schemes with small growers; they select top grade
for export, top-second grade for supermarket channels, and the rest goes to the wholesale
market. This seems a likely scenario over time for supply expansion to Kenyan
supermarkets as medium and large growers need to expand volumes beyond their own-
production capacity.

On the other hand, supermarkets can increase their produce volumes by drawing
on imports, creating intra- and extra-regional trade. Note that Kenyan supermarkets (and
the traditional sector brokers) are already selling cheap produce from the East Africa
region and even cheaper produce from China, such as garlic. What is to prevent Kenyan
supermarkets to increase sourcing cheap produce directly from Tanzania or Uganda in the
region or from more distant, but very price-competitive suppliers in India or China?
Kenya’s leading supermarket chain (largely govemment-owned Uchumi) has always

stuck to its policy of sourcing at least 85% of what is sells from Kenyan suppliers (Daily

61

Nation 2004), but can it afford to continue to do so in the increasingly competitive
supermarket environment? Government policy, such as regional trade policy, will also
affect the extent to which supermarkets will be able to expand produce supply from
external sources or will have to rely on local producers. From a policy vieWpoint, the
issues are mixed. The Tanzanian government reacted to Shoprite importing cheap
tomatoes by imposing protection at the behest of local producers. In contrast, Uchumi
indicated a willingness to source Ugandan produce to send to Kenya as part of its entry
into Uganda. And Uchumi sends Kenyan produce into its Ugandan stores. The
supermarket chains will be motors of trade in the region, and the debate about the pros
and cons of protection and regional trade liberalization will be magnified as the chains

spread.

5.3 Impending Challenges to Smallholders

The above means that small farmers are only one of several options for supermarkets to
turn to as their growth strains current procurement systems and sources. And it means
that supermarkets may join policy and other market-institutional changes in Kenya as key
motors in transforming the food markets facing small farmers.

The squeeze on small producers, as it is manifesting itself, comes through
supermarkets’ reduced reliance on brokers, who source much of their stocks from
smallholder producers. Smallholders are thus squeezed out as indirect suppliers via
brokers to supermarkets, but smallholder supplies can replace the large farmer supplies

diverted in the traditional system. Hence, at this initial stage of supermarket development

62

in Kenya, the rise of supermarkets is not yet excluding small farmers from supplying the
market, but is rather reducing the presence of their large/medium competitors from the
traditional system, opening up an opportunity for them in the traditional system to a large
extent, and in the supermarket market channel to a limited/modest extent at least as long
as supermarkets can keep up their 18% annual growth, while sourcing mainly from
larger/medium commercial farmers.

However, looking to the next 5-10 years, there are several impending challenges
and opportunities that will face small farmers as the supermarket sector develops, and
policy-makers and development programs should observe the situation closely in order to
help farmers adapt to it.

(i) As supermarkets increase their share in urban markets and the traditional retail
market concomitantly shrinks, small farmers, if they want to sell to urban markets, will
increasingly have to sell to supermarkets or the surviving non-superrnarket urban food
market that is competing with them. The latter competition may take the form, as
elsewhere in the world today, of small shops aggregating to form procurement clubs, or
open air or covered markets improving quality and hygienic standards. All this means
that the market requirements facing small farmers will follow an upward trend, whether
directly or indirectly caused by the retail transformation.

This does not mean that small farmers should be aiming at shifting from other
buyers to only selling to supermarkets, even if they are able to do so. It is clear that, with
the combined vicissitudes of any given retail company (such as the trade press currently
reports for Uchumi; Wachira 2004) as a client for a farmer, and the stiff requirements of

participating in the supermarket channel, it behooves a farmer to manage risk, seek

63

market options for the various grades of his/her product, and sell to a portfolio of market
channels, including exports, supermarkets, and wholesale markets, not to mention
keeping the rural market option.

(ii) Much of the direct sourcing from small farmers by the major chains is done at
present in the ‘up-country’ stores that are not — yet — well connected to nationwide or
regional sourcing systems. The development of efficient transport, distribution centers,
and cold chains will, in Kenya as elsewhere, make it easier for the chains to source from
all zones of the country. Thus, current small-farmer suppliers for up-country stores will
increasingly have to compete with medium/large farmers, and competitive small farmers,
throughout Kenya for a number of product markets that were formerly de facto protected
by high transaction costs for the chain, but will soon not be. International procurement
from a number of countries will extend this competitive pressure on the local small
farmer competing with Chinese or Tanzanian farmers in the procurement arena.

(iii) A critical issue is the extent to which small farmers themselves will make the
requisite investments (in risk management, product diversification, value added, and
marketing expansion) to take advantage of the important opportunities provided by the
supermarkets. Much will depend on the terms and conditions they face from the buyers
(the chains or their dedicated wholesalers), such as contracts that reduce risk, limited
payment periods to cut the financial burden, and so on. We have seen in other countries
that small farmers, on the margin of profitability and with few if any cash reserves, are
sensitive to these conditions, which are, of course, also the subject of government
regulation and/or private sector commercial (codes in Europe, the United States and parts

now of Latin America and Asia. Much will also depend on whether governments, NGOs,

64

and donors build up the capacity of small farmers, in particular in groups, to sell to this
channel (in particular, the vector of capital needed to meet the market requirements of the
supermarket channel, including physical, organizational, social, and human capital). This
essay is limited in its scope to data from retailers and wholesalers, but essay 3 presents
detailed data on grower decisions to participate in the supermarket vs. traditional market
channel, and their relative costs and benefits, shedding further light on these emerging

issues of grower participation.

5.4 Implications for Development Programmes and Policy

Given the impending challenges quite probably just around the corner, the next few years
provide the crucial window through which government, NGO, and donor programmes
can help to upgrade the upper tier of small farmers to be able to position themselves as
suppliers to this dynamic urban market. The changes in the procurement systems
discussed above (of supermarkets, and probably also of other urban retailers competing
with them) suggest that this upgrading will need to be diverted towards increases in
quality, consistency, volume, new commercial practices, and use of new technologies.
We have briefly shown that it is possible for small farmers to participate (such as
the case of small farmers selling greens to Uchumi), and we believe that these
experiences can and must be replicated and scaled up. This is now done mainly through
traditional intermediaries who aggregate product over many small farmers and has the
advantage of overcoming the volume and transaction-costs problems facing individual

small farmers selling to supermarkets, but has the disadvantage of relying on a declining

65

market (the traditional broker) and forgoing the margin. Two other options are attractive:
working for a grower/shipper (supplying local supermarkets) in an outgrower scheme
under contract, or supplying directly via an association bulking, grading, packing and
shipping product from its members. Development programs and policy-makers can do
several things to help farmers take, and prosper in, these paths.

First, governments, NGOS, and donors can facilitate small farmers’ access to three
key elements in order to have the capacity to supply the supermarket channel: (i) market
information identifying the buyer and its requirements, and establishing a market
relationship such as having an implicit or explicit contract from the supermarket or the
specialized wholesaler, i.e., being on the list; (ii) viable organization/association to
reduce co-ordination costs and enforce delivery from members; and (iii) the requisite
physical investments (say in equipment) and managerial improvements to meet the
Specific product and transaction standards required by the supermarket chain. A program
that aims to assist a farmer to sell to supermarkets (or outgrower schemes or wholesalers)
by providing one of these but not the others will not work; for example, association-
building is necessary but not sufficient. Berdegué (2001) illustrates this point with
examples of successes and failures of small farmer organizations in Chile. Moreover,
these elements can be mutually reinforcing, for example where having a contract (being
on the preferred suppliers’ list) acts as a substitute for collateral, inducing a bank to make
a loan to a small farmers’ group for the purchase of equipment (an example from Croatia
is given in Dries et al., 2004).

Second, governments, NGOs, and donors can facilitate tri- or quadri-partite

relationships that facilitate smaller farmer participation. An example can be found in

66

Indonesia, where there is a combination of a small farmer organization (Makar Buah), a
supermarket chain (Carrefour), a seed/chemical company (Syngenta), a government
extension program, and a specialized/dedicated wholesaler (Bimandiri), in a fruitful
combination to market melons (Reardon, 2004). Carrefour supplies the guaranteed
market, Syngenta the financing, and the wholesaler the interrnediation and co-ordination.
This kind of combination can also be accomplished through a donor development project,
such as the USAID/Michigan State University Partnerships in Food Industry
Development - Fruits and Vegetables project (PFID-F&V) in Nicaragua, where a US
university facilitates the market connections for small farmer organizations with the local
supermarket chains, NGOs such as Technoserve provide technical assistance,
supermarket chains such as CSU and La Colonia provide the guaranteed market, and
donor funds from USAID provide the financing. The project has a graduation policy
whereby the small farmer-organizations progressively take over the needed investments
and then maintain the market link themselves (Weatherspoon and Membrei’io, 2004).
Third, assistance programs focused on agricultural diversification can help small
farmers with the necessary training and equipment, such as transport and cold chains, to
produce those perishables for which these farmers are likely to be competitive in the
supermarket market-channel. This assistance should be carried out in a way that walks
the tightrope between alleviating poverty as much as possible, and inducing distortions in
market or investment incentives. We have seen above that these products are mainly the
highly perishable products such as leafy greens (e.g., traditional African vegetables) in
which a correctly equipped smallholder can have an advantage. Assistance programs are

likely to find further promising new market opportunities amongst the wide set of

67

produce items for which supermarkets are eager to shift away from the current traditional
brokers to more direct supplies by farmers (e.g., potatoes, carrots). In some cases this will
imply producers from different regions to cooperate with each other in order to address
the year-round delivery requirement of supermarkets currently resolved by brokers (e.g.,
for mangoes). By contrast, we expect rapidly increasing cost and quality competition for
small farmers competing in this channel in bulk products such as bananas and tomatoes.
These can be large-volume profitable enterprises, but the programs should be aware of

the competitiveness bar that will steadily be raised.

68

ESSAY 2
SUPERMARKETS AND CONSUMERS IN AFRICA: THE CASE OF NAIROBI

1. Introduction

Over the past decade, supermarkets have emerged as important agents of change in the
agri-food systems of developing countries (Reardon et al. 2003); for example the share of
supermarkets in domestic retail rose roughly from 15% to 55% on average in Latin
America from 1990 to 2002 (Reardon and Berdegue 2002). This emergence could only
have taken place in the context of changing behavior of consumers who, in liberalized
markets, ultimately decide which retail firms and what products succeed. If the appeal of
supermarkets were limited to the higher income segments of the population, little change
in retail structures would have been noticed. As we show in this essay for the case of
Nairobi, the appeal of supermarkets is much wider with the majority of consumers, not
just the rich, changing their shopping behavior both in response to and driving the growth
of supermarkets.

Agri-food firms, whether domestic or foreign, who want to sell their products in
the emerging markets created by the rise of supermarkets need a good understanding of
the dynamic relationship between the consumer and the supermarket. Not withstanding
this importance, few studies have analyzed the role of consumers in the rise of
supermarkets in emergent markets in detail. Some recent important exceptions include
Zhang (2002) for China, Al-Mazrooei et a1. (2001) for Oman and Rodriguez (2002) for

Argentina. Although supermarkets are also on the rise in Africa in general

69

(Weatherspoon and Reardon 2003), and in Kenya in particular (Essay 1), there have been
no such consumer studies in this region.

By analyzing consumers and supermarkets in the Nairobi food market, we address
this gap in the literature. This exploratory study aims at shedding light, from the
consumer perspective, on the positive feedback loop between consumers giving their
dollar vote to supermarkets and supermarkets using it to create more value for consumers.
More specifically this research analyses the following research questions: (1) which
attributes of retail outlets and the products they sell are important to consumers when
deciding where to buy their food?; and (2) which socio-economic and demographic
factors affect retail outlet choice and shopping frequency? We focus on food in general as
well as on fresh fruits and vegetables (FF V). The latter allows us to assess if consumer
behavior is different between the dry foods and fresh foods categories as food
perishability is likely an important factor in a country like Kenya where few households
haVe refrigeration.

The essay proceeds as follows. The next section describes the data and methods
used in this study. Section 3 describes the rise of supermarkets in Kenya in a broader
context. Section 4 then analyses the patterns and determinants of consumer behavior with
respect to retail outlet choice and shopping frequency. Section 5 provides the conclusion

and implications.

70

2. Data and Methods

This consumer study is part of a broader study on the rise of supermarkets in Kenya
which included primary data collection during a lO-month fieldwork period (March to
November 2003, April 2004). The consumer study had two parts.

The first part in the consumer study consisted of consumer focus group research.
Eight focus group sessions were conducted in Nairobi. These included four all-female
and four all-male sessions with a total of 54 participants from all income categories.
Following standard guidelines for focus group research (see e.g., Churchill 1999), we
controlled each session for gender and income class so that within a specific session
participants were similar in those two respects and they would feel at ease
communicating in the group. These focus groups allowed us to get familiar with nature of
and the main motivations behind the shopping behavior of Nairobi’s citizens with respect
to food in general and fresh fruits and vegetables in particular. It also allowed us to
understand the vocabulary used by consumers to describe their shopping behavior, thus
further facilitating a good design of the survey instrument.

The second part consisted of a consumer survey. The designed survey instrument
focused on two sets of comparable questions, one for food and one for FF V. These
question sets related to retail outlet choice, shopping frequency, expenditure levels and
reasons for retail outlet choice. A set of socio-economic questions were included to
obtain the independent variable values. This instrument went through several pre-tests

and revisions before a final format was decided upon.

71

In order to get a good geographic and income-class representation, a stepwise
sampling process was implemented. Within each of Nairobi’s eight administrative
divisions (used as sub-population strata), we selected (within the context of the stratum)
the lowest, a middle- and the highest income location. This selection was based on the
poverty gap for the location (CBS 2003). The poverty gap measures how much those who
are under the poverty line are, on average, below the poverty line. For each selected
location, we then conducted a field reconnaissance with the assistance of the location
Head (Chief) and composed a list of the estates (i.e., residential neighborhoods
demarcated by a high degree of internal socio-demographic Similarity). From this list we
then made a random selection of 24 estates. Enumerators were assigned a street (usually
the main street) in an estate to start from and were free to select a (any) dwelling in the
street as the starting point. Subsequent interviews then took place at each 3rd house until a
pre-deterrnined number of interviews, relative to the population size of the location, was
attained. A similar sampling design was used by Zhang (2002) in China. If a household
refused to collaborate or was not home, it was skipped. In order to minimize this
replacement method leading to a bias against small households (where all family
members work during the day on weekdays), interviews were conducted in the weekend.

A total of 445 valid interviews were conducted in the interviewees’ homes during
November 2003 by a team of 18 business students from Kenyatta University.
Enumerators worked in several rounds of data gathering so that their work could be
checked and they could be re-instructed were needed as the survey progressed. Response
rates were high, except in the high income estates where enumerators indicated that it

was very difficult to get admitted by the house-guards to households. (Zhang (2002)

72

made a similar observation.) This led to an under-representation of the high income
households in our sample. In order to correct for this, weights were used where the
analysis below describes the population as a whole. These weights are based on a
distribution of Nairobi’s population over different income-classes estimated from various

sources (World Bank 2004, CBS 2002a, Obudho 1997).

3. Supermarkets, Population Dynamics and the Urban Food Market in Kenya

It is important to first give the definition of a supermarket we used in this study as this
definition varies from country to country. The terms “self-service store” and
“supermarket” are used interchangeably in Kenya, irrespective of their size, as both have
exploded on the retail scene together. Furthermore, supermarket chains have developed
different store formats to reach different customer segments. Hypermarkets with large
parking areas along the main entry/exit roads in Nairobi and Mombasa mostly attract the
high and middle-income consumers. Smaller ‘neighborhood’ stores in Nairobi’s
residential areas (‘estates’) mostly target middle-income consumers. City center stores
near busy bus stages mostly attract the middle to low income consumers. Based on the
Kenya branch of the international retail auditor AC Nielsen, supermarkets are here
defined as “self-service stores handling predominantly food and drug fast moving
consumer goods (FMCG) with at least 150m2 (1,625 sq.ft) of floor space”. We defined
supermarkets of 15 times this size as hyperrnarkets (i.e., 2,250m2 or 24,460 sq.ft). Using

these definitions, there are about 209 supermarkets and 16 hyperrnarkets in Kenya.

73

At the same time, an estimated 900 to 1,400 smaller self-service shops have
entered the retail sector. These shops include mini-supennarkets in smaller towns as well
as convenience stores in residential areas and at gas stations. In small towns, the
emergence of these mini-supermarkets is just as radical a departure from the traditional
shopping experience of consumers as hyperrnarkets are to consumers in Nairobi.
Although this implies that small self-service shops are an important part of the retail
revolution taking place in developing countries, this essay focuses on the 225
supermarkets meeting the size criteria above (essay 1).

In 2003, supermarkets sold US$520 million worth of products of which US$365
million in food — or roughly 20% of the US$19 billion urban food market in Kenya
(essay 1). The rest of the urban food market consists of smaller self-service shops which
represent 17% of the urban food market, while the remaining 63% of sales comes from
traditional retailers. Since 1995, supermarkets (in terms of their aggregated sales) have
been growing at an average annual growth rate of 18%, i.e., much faster than aggregated
urban income. Consequently supermarkets are increasing their market share vis-a-vis
traditional food retailers such as kiosks, greengrocers, over-the-counter shops, market
stalls and street hawkers and at current growth rates, supermarkets will become the
dominant food retailers by 2011.

The rapid grth of the urban population, due to both overall population growth
and urbanization, has been a key driver of the growth of supermarkets since their

emergence in the 19605”. Supermarkets are located in urban areas only in Kenya, hence

 

'2 The urban population is here defined as all people living in urban centers of 10,000 or larger. Based on
the 1999 population census (CBS 2002b) and UN growth projections (UN Habitat 2004, UN Population
Division 2003) we estimate the urban population in 2002 at 11.9 million or 37% of the overall population.

74

our focus on urban consumers. Between 1989 and 2002, Kenya’s population has grown
from 21 million to 33 million and is expected to reach 42 million by 2010 (Mungai et a1.
2000). In addition, the average urban population growth rate is double that of the overall
population grth rate (United Nations, 2002). In 2003, nearly 40% of Kenya’s
population and half Kenya’s households (given smaller household sizes in cities) already
lived in urban centers and the rural population, in absolute numbers, has started to
decrease. Given currently anticipated growth rates, the urban population is expected to
surpass the rural population by 2013. Although the overwhelming majority of the people
migrating to and living in Kenya’s urban areas are poor, their aggregated demand
constitutes the effective demand modern supermarkets, with their low margin - high stock
turnover strategy, can thrive on (Pralahad and Hammond 2002).

The effect of urbanization is amplified by a related change in life-style. Relative
to rural households, urban households have less time to shop for food and prepare meals
(e.g., because of women working out of the home or men working a migrant labor to
provide for their rural family), have higher incomes and are more mobile. At the same
time, on-going market and trade liberalization since the mid-19905 increased product
variety in the Kenyan market place dramatically and induced a shift from a supplier to a
consumer-driven economy. Increasingly, supermarkets, with their broader assortment
provided an alternative for traditional retail outlets with limited assortrnents. The urban
population also consists overwhelmingly out of young adults (43% are aged between 15
and 34, CB82002a) who have embraced a westemized life-style in which supermarkets

are the retailer of choice.

75

Fresh fruits and vegetables (FFV) is a fairly new and, given the high perishability,
far more difficult to manage product category for supermarkets in Kenya. Nevertheless,
FFV hold great potential for supermarkets looking for growth opportmrities because they
are an important part of the diet of urban consumers in Kenya who (for vegetables)
consume twice the volume of rural consumers (i.e., an estimated annual per capita
consumption of 40kg; GoK 1994) and, as they generally have less time and land to
produce their own fruits and vegetables, spend 7% of their income on buying fi'uits and
vegetables from urban retailers (CBS 2002a). Although the supermarkets only had a 4%
share of the US$364 million urban FF V market in 2003 (that share is 6% in Nairobi), on-
going changes in their procurement system in combination with changing consumer
habits are expected to increase that share substantially over time (essay 1). While selling
fresh fruits and vegetables in supermarkets in not a rarity, given that overall four out of
ten supermarkets sell them to various degrees, it is very much concentrated in the Nairobi
hyperrnarkets of two leading supermarket chains. Produce sections in these hyperrnarkets
(which attract the high-income consumers) can take up 7% of the store’s sales floor space
and offer over 300 stock keeping units (SKUS) from American pink sweet potatoes to
zebra yellow melons (although only 100 or so are available at any given time)”.

Kenya’s supermarket sector has three tiers (essay 1). The first tier consists of the
two clear market leaders and also the leading produce retailers, Uchumi Supermarkets
and Nakumatt. These two supermarket chains represent nearly 50% of the supermarket
sector in terms of sales. While Uchumi targets consumers from all socio-economic

classes, Nakumatt’s consumer focus has been mostly on the high-income segment (50%

 

‘3 For comparison, fresh produce sections in US supermarket chains have on average 634 SKUS (Perosio et
a1. 2001).

76

of their customers fall in this category). The second tier consists of the Tusker, Ukwala
and Metro Cash ‘n Carry chains. These three chains are fiercely competing for the
shilling vote of the middle—to-low-income urban consumers. The Top 5 supermarket
chains represent 28% of stores and 60% of the sales, indicating a concentrated sector. The
supermarkets in the third tier exist of small chains and independent (single-store)
supermarkets.

In terms of geographic spread, supermarkets have expanded rapidly although the
consensus in the sector is that there still remains much room for growth. In 2003, nearly
60% of the supermarket outlets (and 44% of the supermarket sales) were located outside
of Nairobi and essentially every provincial capital in Kenya had one or more
supermarkets. However, the top five chains are present in only eight urban centers and
only about half the urban centers with a population of more than 25,000 have a
supermarket. The country-level supermarket density of 7 supermarkets per million people
is still far below that of countries with similar incomes and urbanization rates (such as
Central America with double the rate).

The Nairobi food market is an interesting case for several reasons. With a city
population of about 2.6 million, it is the largest city in East Africa (and the 12‘h largest in
Afiica). Its population is growing fast, with a recent study predicting that (mainly through
rural-to-urban migration) Nairobi’s population will grow to 7.5 million over the next 20
years, i.e. an increase of almost 700 additional inhabitants every single day (African
Population and Health Research Centre 2003). Nairobi also leads the rest of the country
in several other aspects: (1) Nairobi still accounts for majority (56%) of supermarket

sales; (2) in terms of store density, Nairobi is clearly in the lead with 36 supermarkets per

77

million, nearly triple the level of the rest of urban Kenya (comparable to a Central
American city, Berdegue et al. 2004); (3) the FFV market share of supermarkets is also

the highest in Nairobi (at 6%).

4. Empirical Results and Discussion

4.1 Supermarket Customers and Household Income
Table 6 reports results from the Nairobi consumer survey. In Nairobi, 80% of households
buy part of their food from supermarkets on a regular basis, i.e., at least once a month
(Table 6). That figure is 60% for the lowest income-group. Households with monthly
incomes of less than Ksh15,000 make up 56% of the customers in supermarkets and 36%
of their sales. Table 6 also illustrates why the upper-middle and high income classes are a
key battleground for the leading chains: although they make up only 15% of the
population they represent 44% of the supermarkets’ sales. At country level, given the far
higher average household income in Nairobi (CBS2002a), low-income consumers are
expected to be even more important in terms of the percentage of customers and revenues
they represent with regard to supermarkets.

The above findings allow us to estimate the importance of the relatively large
high-income Asian Kenyan and Western expatriate segment of the population (about
200,000 in 2003, i.e., 1.7% of the urban population)“. While this segment has played an

important part in providing a nucleus from which supermarkets have been able to arise

 

‘4 This number is estimated by the author from data on the ethnic distribution of the Kenyan residents
(based on the 1999 population census) as received from Corporate Africa (a Nairobi-based frnancial
services frrm that also specializes in entrepreneurship education and development, policy analysis and
research).

78

.383 .8835 8.8.85 75 v5 A333 «853.— 75 A335 sea—>5 cone—Eon— 75 A338 .8880

283 3cm Siam 233 038 988 ducammmo .858 88 .0888 wage—cu 05 :o «.83 8e 88. :93:qu $3.5m 888:8 m 558 .8858
Eco 02886:. 038 05 E 8% 05 83E :6 bowofio 088:. aw:— ofi E 8:93:88 mo

c383: 3o. 05 98 88.88.88 8 30% 8:68.28 8:88 .«o mac—£5 882888 8836288 “.83- :82 05 0885 Co 838 8:88.28 95 .8oz

 

 

 

$2: $2: $e $2 ..\..8 83.25." $2: =<

$mm $N~ $3 $3 $3A 836m— $e 3862an A :3:

as s: :2 sea :8 838 .8 832-8%: 822 can:

$8 $NN $m $2 $3 3ch $3 38de Emu 0:922 83o;

$3 $2” $m $m _ $3 3863 $3 89m 723— 33

$2 $3 $_v $_v $8 38.22. $3 38.23. V Loom
820 So...— 880 EEG .oJEanA—am «excaflconsm «exuanasw 880 3.5

83m 82:82.0 5 i.e.—m E 825...— 5 ”5.525 5 9.98m 23 8:8:— 880
«Stageham nexus—Eonsm 85:327..“— ..8....~ 9.3.5 . . . I: >328: «E8:—

.e .x. e. .\. E... a: he .\. 3.6 e. .x. 3.6 .e .x. ._e a co .\.

mean i.e.—HZ E nae—U 2:8..— .3 98m 8589.50 .floxuufiuomam 2:. 5 93:.

79

(and these segments are still an important part of the customer base, especially for fresh
produce), supermarkets have moved well beyond this segment. In 2003, the Asian and
expatriate community, which made up an estimated one third of the upper-middle and
high income consumers, represented only 15-20% of supermarket sales.

The Nairobi consumer survey also revealed a sharply different picture with regard
to the buying of fresh fruits and vegetables. Only 15% of Nairobi’s households get (some
of) their FFV from supermarkets. Less then 1% of the households with incomes of less
than Ksh5,000 buy fresh produce from supermarkets. Only 15% of the households in the
next income-classes (Ksh5,000 to Ksh40,000) buy fresh produce from supermarkets.
Because these 15% do not buy all of their fruits and vegetables from the supermarket,
only 5% of the total FFV expenditure of households in this income class is spend in
supermarkets. Buying F FV from supermarkets becomes more important for the upper-
middle and especially the high-income households of which 30% and 67% buy FFV in
supermarkets respectively. However, even in the high income class, only 40% of the FFV
expenditure goes to supermarkets. In other words, supermarkets do not dominate FFV
markets in any income-class at this point in time, although they are getting close with
regard to the highest income class.

How can we explain these patterns? Our consumer focus group research in
Nairobi’s low-income neighborhoods (e.g., Kibera, one of Africa’s largest urban slum
areas) indicates that most of the residents shop at nearby supermarkets, albeit not so
frequent (mostly once a month) and for small values at a time. The low per capita
expenditure of low-income consumers is partially off-set by their vast numbers (e. g.,

Kibera alone has an estimated 800,000 residents). Table 7 ranks the reasons why

80

consumers in Nairobi buy food from supermarkets or from other retail outlets such as
kiosks or over-the-counter shops. Traditional shops hold two key advantages over
supermarkets, especially for the poor: (1) they are easy to get to, and (2) they provide
credit. Especially the convenient location of kiosks, right next to where consumers live, is
important as the majority of consumers (70%) shop from home rather than from work.
Nearly 90% of those who shop in traditional shops cite location as a reason to shop at
traditional retailers, while for nearly 60% of them it is the number one choice criterion.
Credit is an important strategic advantage that traditional shops hold over supermarkets

because it implies a long term relationship: kiosks or over-the-counter shops provide

credit to their regular customers.

Table 7: Reasons for Retail Outlet Choice in Food by Nairobi Consumers

 

 

 

 

 

Reasons to Buy Food Reasons to Buy Food
from Supermarkets (N=338) From Other Retail Outlets (N=335)
% of % of 4 % of % of
Customers Customers Customers Customers
Reason Ranking it indicating Reason Ranking it indicating
as the No. it as a as the No. it is a
1 reason reason 1 reason reason
Low Prices 58% 81% Easy to Get To 58% 87%
Large Assortment 23% 64% Credit Available 17% 49%
Easy to Get To 8% 45% Low Prices 12% 25%
Product Quality 4% 22% Friendly Service 2% 12%
Easy to Buy Bulk 3% 23% Packaging 2% 6%
Other Reason 4% Na Other Reason 9% Na
100% 100%

 

Source: author’s consumer survey (Nairobi, November 2003).

The two most important incentives for consumers to shop at supermarkets are the lower
prices and the large assortment. Table 8 compares prices for a selection of high-turnover
processed food items between supermarkets and traditional food retail outlets. Prices in

supermarkets are generally lower, albeit with only 3-4% on average. Nevertheless, as in

81

other countries, low prices are single-most important driver behind the fast consumer
acceptance of supermarkets in Kenya, especially amongst the more price-sensitive low-
income consumers. From the focus group participants in low-income neighborhoods in
Nairobi we further learned that they buy mostly easy-toastore bulk goods such as a 2kg
bag of sugar or a rod of soap from the supermarket, while for their smaller volume
purchases (e.g., SOgr of loose sugar) they go to kiosks who can sell from bulk bags in any

quantity the customer wants (or can afford!)

Table 8: Supermarkets vs. Other Retailers Price Comparison for Processed Foods

 

Selected Basic Processed Foods

Price Price at Price Difference with Nearest Traditional Retailer:

 

 

at Nearest
Item 323;; R'le‘gi‘ller Alflasrlkgg- Ukwala Uchumi Nakumatt Tusker
(Ksh) (Kali)

Loose Ie_a 100g:- 31 32 -4.8% -6. 1% -4.9% -3 .8% -0.6%
M 2kg 99 104 -4.7% -9.0% -6.4% -3.2% -3.3%
Maize Flour 2kg 52 54 -3.7% -5.2% -2.0% +0.7% -0.8%
O_il 1kg 95 97 -2.7% -4.3% -2.8% 4.1% -1.8%
Wheat Flour 2kg 64 65 -2.5% -6.3% +0.3% -2.9% +1 .6%
Bread 400gr 20 21 -1.9% -3.9% -0.8% -0. 1% -l .0%
All 6 Items 361 373 -3.4% -5.8% -2.8% -2.3% -1.0%

 

Source: author’s national supermarket survey (May-July 2003). Traditional retailers are kiosks or over-
the-counter groceries.

In F F V retailing, supermarkets have not penetrated the middle and lower income classes
to the same extent as they have for food. Table 9 ranks the reasons why consumers in
Nairobi buy FFV from traditional shops or supermarkets. Consumers (especially the
poor) continue to buy FFV from traditional retailers (mostly open air markets, kiosks and

greengrocers) for three main reasons: (1) price, (2) retailer location, and (3) freshness.

82

Table 9: Reasons for Retail Outlet Choice in Fresh Produce by Nairobi Consumers

 

 

 

 

 

Reasons to Buy Fresh Produce Reasons to Buy Fresh Produce
from Sugrmarkets (N=77) From Other Retail Outlets (N=397)
% of % of % of % of
Customers Customers Customers Customers
Reason Ranking it indicating Reason Ranking it indicating
as the No. it as a as the No. it is a
1 reason reason 1 reason reason
Freshness 27% 44% Low price 38% 58%
Quality 18% 3 1% Location 29% 57%
Assortment 1 7% 3 5% Freshness 1 7% 5 1%
Low price 12% 31% Assortment 6% 20%
Location 9% 24% Friendly Service 3% 14%
Other 17% Na Other 7% Na
100% 100%

 

Source: author’s consumer survey (Nairobi, November 2003).

Unlike for dry foods, prices for FF V are generally higher in supermarkets because (1) the
produce items have a higher value-added (e.g., they are washed, sorted by size and/or
quality grade, and so on) and (2) supermarkets apply a higher mark-up to their purchase
price in part to cover higher overhead costs relative to traditional retailers (e.g., facility
rental costs). For a selection produce items, table 10 compares the price at the
supermarket with those at traditional FFV retail outlets. These price comparisons have an
important limitation as they do not take differences in quality into account (the quality
being generally more consistently higher at supermarkets and greengrocers than at open
air markets). With this limitation in mind, two key findings emerge from table 10. First,
supermarkets are close to becoming price-competitive with the greengrocers, kiosks and
covered market stalls where the middle-to-high income groups traditionally bought their
FFV. This implies that with little difference in quality and prices, greengrocers, kiosks
and covered markets catering to the higher income groups have been the hardest hit by
the introduction of FF V in supermarkets as the latter hold the strategic advantage of being

able to offer the convenience of one-stop shopping to their customers.

83

Table 10: Supermarkets vs. Other Retailers Price Comparison for Fresh Produce

Selected Fresh Fruits & Vegetables
Price Difference (negative. % = lower price in

 

 

 

 

Average Price
at so ermarket . .
Item Supermarket Covered GAPS," , Greengrocer “$332311
"(sh/kg) Markets Markets residential area)

Spinach 12 -36% -5% -22% -31%
Kale l 5 -14% +49% -8% -33%
Oranges (local) 43 -10% +88% -9% -18%
Cabbage (green) 12 -9% +89% 0% -28%
Carrots 24 +1% +95% +30% -35%
Tomatoes 44 +14% +52% +10% +8%

Mango (N gowe) 56 +17% +129% +22% -5%

Potatoes (Meru) 27 +20% +160% n.a. +8%
Bananas (eflg) 50 +41% +108% +7% +71%
All 9 Items 283 +6% +87% +24% -1%

 

Note: if prices were not indicated in Ksh per kg, the produce was bought and weighed. Were possible
prices were negotiated as to reflect normal buying patterns. Prices reflect differences in quality, sorting
and value-adding (e.g., cleaning) which could not be taken into account.

Source: prices for fruits and vegetables were collected in Nairobi on August 14 and November 14, 2003
and April 14, 2004 at the Jogoo Road, Railway and Westlands branches of Uchumi, the covered markets
Westlands City Council Market and City Park Market, the open air markets of Wakulima and Kangemi,
Zuchini (a high-turnover greengrocer frequented by high-income consumers), and a group of kiosks in
Highridge, a high-income residential area. Zuchini, City Park and kiosk prices were collected on April 14,
2004 only.

Second, supermarkets are, relative to the open air markets where the urban poor buy their
FFV, substantially higher priced (on average nearly 90%) for most (but not all) produce
items. The leading supermarkets chains, who represent 90% of the FFV sold through
supermarkets and who did not start selling FF V until recently (Uchumi since 1997,
Nakumatt since 2001), at first focused on getting the quality right for the (less price-
sensitive) high-income consumers they were targeting. Nevertheless, table 10 shows that
price differences are becoming smaller for some key products (tomato, kale) while for
spinach supermarkets are already selling at a lower price than any other retailer in
Nairobi. Another indication that supermarkets are bridging the price-differential is given
by the observation that street hawkers and small shops have started sourcing their F FV

from supermarkets like Uchumi (especially in middle-income residential areas). These

84

traditional retailers make their margin by dividing up in smaller quantities and by
bringing the produce closer to the consumer (but at a higher price per kg).

Apart from price, retailer location plays a key role in the consumer’s decision on
where to buy FFV. Even if supermarkets can bring theiriFFV prices down to the level of
the open air markets and the kiosks in the low income neighborhoods, it is unlikely that
this will attract large numbers of poor consumers. Because low-income household do not
have refrigerators, they are forced to buy perishables (like milk and vegetables) in small
quantities at higher frequency (usually daily)15 . From the focus groups we learned that
poor households work with a more or less fixed budget for FFV, e. g., KshSO per day. The
ubiquitous kiosks are usually within a few minutes walking distance, while getting to the
far less omni-present supermarkets (especially those selling FFV) often requires a long
walk or maybe even a bus ride. Taking a bus for KshS only to buy FFV for Ksh50 will
make little economic sense to consumers, especially if it concerns a daily shopping
activity. In other countries, supermarket chains developed smaller store formats that can
more easily be integrated in residential areas (e.g., superettes in South Afi'ica). Such store
formats have not yet sprung up in Kenya (Metro’s Lucky 7 trading group comes close but
does not yet distribute FFV). However, the smaller supermarkets that Uchumi opened in
Nairobi’s middle income residential neighborhoods (e.g., Buru Buru) have been very
successful in selling FFV. This suggests that the location card may play much less against
the supermarket chains in the future in produce retailing.

Table 9 lists freshness as the third key reason for consumers to select traditional

retail outlets as their source of FFV. When supermarkets just started selling FFV, they

 

'5 Zhang (2002) found similar results for the case of Shanghai, China.

85

had to go through a learning process with regard to procurement and shelf management.
In those early days, shelf management was not optimal in the produce section. Supplies
were late, leading to empty shelves or produce was kept on the shelves for too long,
leading to an unattractive presentation (less bountiful, less fresh). In contrast, traditional
retailers have over the years perfected the art of presenting FF V at their freshest. This
involves excellent matching of supply with demand as well as being well-linked to other
retailers (most notably street hawkers) to whom less fresh produce could be sold at
starkly reduced prices (but not at a complete loss). This phase created a perception in the
consumer’s mind that FFV sold in supermarkets are less fresh. Although supermarkets
only buy the best quality FFV available (in the volumes they need) at any given time
(varies throughout the year), stock-outs and keeping the produce too long on the shelves
still occurs sporadically (even for high turnover items like bananas).

Nevertheless, the findings of this study reveal that Nairobi consumers feel that
supermarkets have made great progress, since freshness is the most frequently no. 1
ranked reason for consumers’ buying FFV from supermarkets. The consumer focus
groups further revealed that low-income consumers mistrust the presence of refrigeration
which they believe allows supermarkets to keep produce beyond their ideal freshness.
Traditional retailers, who for the greater part have no refrigeration, cannot ‘fool the
consumer’ because consumers would immediately see it if a retailer kept produce for too

long.

86

4.2 Retail Outlet Choice and Shopping Frequency

In this section we want to assess how household socio-economic factors (such as income,
household composition, education, whether there is a maid or not, and so on) influence
the following four decisions: (1) whether or not households shop for food at supermarkets
at least once a month (Y1); (2) whether or not households buy F F V at supermarkets at
least once a month (Y2); (3) how frequently households shop for food at supermarkets
(Y3); and (4) how frequently households shop for FFV at supermarkets (Y4). Since
negative dependent variables are not possible and assuming non-linear effects of the
independent variables, we modeled the first two decisions as probit models for a binary
response (Wooldridge 2000), and the latter two as ordered multinomial logit models with
three frequency levels in the dependent variable, 3=high (more than twice a month),

2=low (once or twice a month) and l=rarely or never (less than once a month) (Greene

2000).

Model (1)

Model (2)

Model (3)

Model (4)

Prob (Y1=1|X) = G(Bon + X131)

Prob (Y2=1|X) = 6([302 + X132)

Prob (Y3=1 IX) = A(-XB3)

Prob (Y3=2 IX) = A(K32-XB3) - A(K31'XB3)
Prob (Y3=3 IX) = 1 - A(K32-XB3)

Prob (Y4=1 |X) = A(-X4)

Prob (Y4=2 IX) = A(K42-XB4) - A(K41'XB4)

Prob (Y4=3 IX) = l - A(K2-XB4)

87

The determinant variables (X) included in this analysis here are: (a) whether or not the
household owns a motorcycle or car (own_tran); (b) whether or not the household owns a
refiigerator (own_refr); (c) whether or not the household has a credit card (has_card); ((1)
whether or not the household has a maid (has_maid); (e) the per capita income of the
household based on income category (8 categories) and household size (pcincome); (t)
the education level of the household heads who do the actual shopping (averaged if both
wife and husband shop) using four categories: O=none, 1=primary, 2=secondary,
3=university (educat); (g) the proportion of young family members (under 30) in the
household (prpyoung); (h) the age category (6 categories) of the responding household
head (agehhead). These variables capture mobility (car or motorcycle ownership), ability
to buy larger quantities of fresh food (refi'igerator ownership), purchasing power (credit
card ownership, per capita income), opportunity cost of time (having a maid or not,
demographic variables) and preferences and life-style (education level, demographic
variables).

Based on maximum likelihood estimation (MLE), tables 11 and 12 present the
probit and logit estimators (,6 i) of the different models. Table 11 indicates that the results

are as hypothesized, namely, the probability of a household buying food from a
supermarket increases as the household has (i) a higher purchasing power, (ii) a higher
level of education of the heads of the household, (iii) younger heads of household, (iv) a
refi'igerator. The impact of car or motorcycle ownership, credit card ownership, whether
or not the household has a maid and the proportion of younger household members did
not have a significant impact on the probability of the household buying food in the

supermarket.

88

Table 11: Probit Results Retail Outlet Choice Consumers

 

 

 

 

 

Model 1 Model 2
P(Shops at Marginal P(Buys FFV in Marginal
Independent variables supermarkets) effect supermarkets) effect
(”IL“) (t-wLue)
Owning car or motorcycle 0.277 (0.88) - -0.016 (0.06) -
Owning refrigerator 0.571 (187*) 10.7% 1.20 (5.04***) 30.5%
Owning credit card -0.l65 (0.54) - 0.201 (0.74) -
Maid help 0.129 (0.53) - -0.1 14 (0.46) -
Income per capita 0.384 (2.04") 8.6% 0.171 (1.51) -
Education 0.242 (2.08") 5.4% 0.216 (1.91 *) 3.8%
Young members proportion 0.064 (0.17) - 0.881 (1.94“) 15.3%
Age of family head -0.169 (1 .81*) 3.8% -0.025 (0.23) -
Constant 0.474 (0.71) - -2.87 (3.57) -
No. of observations 355 354
( Pseudo ) R-sguare 0.095 0.202

 

Notes: * = significant at the 10% level, " = significant at the 5% level, "‘ significant at the 1% level.
Marginal effect measured at the mean levels of the determinant variables.

Source: author’s consumer survey.

Table 12: Ordinal Multinomial Logit Results Shopping Frequency Consumers

 

 

 

 

 

Model 3 Model 4
P(Shopping absolute P(Buying FFV absolute
. frequency in t value frequency in t value
Independent variables supermarkets) supermarkets)
(s.eL (s.e.)

Owning car or motorcycle 0.361 (0.324) 1.12 0.171 (0.432) 0.40
Owning refrigerator 0.108 (0.32) 337*“ 2.25 (0.425) 529*”
Owning credit card -0.983 (0.376) 2.61 ** 0.261 (0.469) 0.56
Maid help -0.027 (0.298) 0.09 -O.248 (0.442) 0.56
Income per capita 0.134 (0.153) 0.88 0.106 (0.16) 0.66
Education 0.283 (0.135) 2.10* 0.339 (0.199) 170*
Young members proportion 0.669 (0.475) 1.41 1.288 (0.785) 1.64
Age of family head -0.216 (0.128) 168* -0.151 (0.199) 0.76
No. of observations 320 354
(Pseudo) R-sguare 0.0510 0.1597

 

Notes: ‘ = significant at the 10% level, “ = significant at the 5% level, "‘ significant at the 1% level.

Source: author’s consumer survey.

When we look at the probability of a household buying FFV in a supermarket (table 11),

a similar picture emerges with, however, the following three differences. First, income

per capita is no longer a significant determinant, which is an unexpected result given that

table 6 indicated a positive correlation between income and buying FFV from the

supermarket. Although we could not identify exactly why, this may in part be explained

89

by realizing that (l) FFV are relatively inexpensive purchases and (2) for some FFV
items supermarkets are price and quality competitive as shown above. Second, the
proportion of younger household members becomes a significant and positively
correlated determinant. This fits the predicted pattern that younger households are
increasingly accustomed to finding (and buying) FFV in supermarkets and consequently
are less conservative in their buying behavior with regard to fresh categories such as
FFV. The third difference is the starkly increased significance of refrigerator ownership
as a determinant of the household’s decision to buy FFV from the supermarket. The
marginal effect indicates that, for the average household (in terms of the independent
variables), having a refrigerator increases the probability that the household buys FFV at
the supermarket by 31%.

Shopping frequency at supermarkets increases as refrigerator ownership and
education increase, as the (responding) household head is younger and when the
household does not have a credit card (table 12). While the first three signs are as
hypothesized, the negative relationship between credit card ownership and the frequency
of shopping for food is maybe less evident, given that the use of credit cards is
encouraged by retailers exactly to smoothen out once-a-month purchases linked to salary
payments. One possible explanation is that credit cards represent a greater flexibility in
purchasing power, and so the consumer is less likely to limit her or his purchases at any
given trip to the supermarket, making additional trips less necessary. Another related
explanation is that in Nairobi credit cards can only be used in the larger supermarkets.

Because these usually require more time to reach and carry larger assortments, credit

90

cards will lead to consumers going less often to make larger purchases from
supermarkets.

Looking across the four models we see that education and refrigerator ownership
are consistently significant determinants in the shopping behavior of Nairobi’s
households with regard to supermarkets. Especially refrigerator ownership, and the
storage capacity it represents, appears to be the key driver. Furthermore, in three of the
four models, younger households or households with younger decision makers have an
increased probability of shopping at supermarkets as well as of doing so more often.
Given the recent nature of the strong emergence of supermarkets and produce sections in
them (i.e., since the mid-19903, less than a generation ago), it is more likely that this
represents a behavioral shift from one generation to the next rather than a decreasing
interest in supermarkets on the part of consumers as they grow older. The focus group
finding that for households with children, shopping at the supermarket is seen as an

outing for the whole family further strengthens this theory.
5. Conclusions and Implications

This essay has demonstrated that supermarket growth in Kenya is following a pattern
similar to that observed in other countries (e.g., Central America). By developing various
new store formats and entering smaller towns and residential neighborhoods,
supermarkets have rapidly increased their share of the urban food market from a small
niche a decade ago to 20% in 2003. Our findings indicate that this market penetration

first occurs for processed and dry foods with penetration in the FFV market lagging

91

behind. Very rarely in the recent work on supermarket diffusion in developing countries
has consumer-side evidence been brought to bear to explain the variations in trends. The
findings here target that gap in the knowledge. Several main findings emerged.

The key finding is that supermarkets are not merely are place where the rich buy
their food. Supermarkets have penetrated the markets for the poor: 60% of Nairobi’s poor
buy some of their food from supermarkets each month while the poorest two thirds of the
population already represent 56% of the supermarkets’ customer base and 36% of their
food sales.

The main reason why the poor buy food from supermarkets is that they are
perceived as being cheaper than the traditional retail outlets, at least for dry foods which
these consumers usually buy once a month. Supermarkets derive this competitive
advantage from the large volumes they trade, which allows to them to buy direct from
manufacturers at discounted prices. For fresh food items, such as FFV, poor consumers
still rely heavily on the traditional retail outlets for two main reasons. First, the poor have
no refrigeration and therefore need to buy these items in small quantities on a daily basis,
making the closer-to-home traditional retail outlets far more convenient. Secondly, FFV
are generally more expensive in supermarkets. However, there are signs that the latter
may change soon. On the one hand, supermarkets are developing smaller retail formats
which can be located closer to where the consumers live. These new store formats may be
the precursors in Kenya of the smaller self-service store chains that have succeeded in
making in-roads in the low income consumer segment in other developing countries, such
as the “superettes” in South Africa (Weatherspoon and Reardon 2002) and the 7-11 stores

in Thailand (Feeny et a1. 1996). These new stores in residential areas not only provide

92

easier access to end-consumers directly but also indirectly through the smaller, traditional
retailers (such as street hawkers) who procure from them (rather than from the farther
away wholesale markets). On the other hand, for some FFV items (most notably leafy
greens such as spinach and kale) supermarkets are already the price leader. It is likely that
more of their FFV items will become price-competitive as supermarkets further increase
the efficiency of their FF V procurement system and pass these savings on to their
customers in a bid to gain market share.

A number of managerial implications for farms and food firms arise from the
foregoing findings. On the one hand, supermarkets create new market opportunities
relative to traditional markets. Supermarkets play into the desire of consumers to have a
broad assortment of products to choose from (consumerism). Increasingly, for both dry
and fresh foods, these are value-added and branded or quality-guaranteed products. The
research here further shows that households that have more younger members or younger
household heads and that have better educated household heads, are more likely to shop
in supermarkets and to do so more frequently. These are households whose shopping
behavior is more likely to be influenced by new product introduction or enhanced product
attributes, such as food safety standards in the case of FF V. As supermarkets further
penetrate the large low income consumer segment, they offer opportunities for their
suppliers to develop and market more products in small-volume, low price packs that fit
into the fixed-budget driven consumption habits of the lower income consumers.

On the other hand, supermarkets bring additional challenges for their suppliers.
Higher quality standards and tougher delivery terms (short order cycle, volume

consistency, supplier credit, and so on) must be reconciled with a competitive pressure to

93

constantly lower costs. Supermarkets further intensify competition between agri-food
suppliers by providing an attractive, low transaction cost point of entry for foreign
producers of agri-food products. For domestic agri-food firms (planning on) supplying
supermarkets this almost inevitably implies the need for an increase in the scale and
sophistication of operations (production and distribution logistics). This essay has shown
that this challenge is coming rapidly for processed foods as this market is becoming
dominated by the supermarkets. The fresh produce growers have a good deal more time
to make the adjustment - and there is time but also incentive for development programs
to either help farmers prepare for the challenges of the supermarket channel, or work hard

to develop alternatives for them.

94

ESSAY 3

FARM-LEVEL PERSPECTIVES
ON THE IMPACT OF DOMESTIC SUPERMARKETS
ON KENYA’S FRESH FRUITS AND VEGETABLES SUPPLY SYSTEM

1. Introduction

The rise of supermarkets in developing countries has received considerable attention in
the development economics literature over the last few years (Reardon and Berdegue
2002, Weatherspoon and Reardon 2003, Hu et a1. 2004). The research presented in those
articles mostly focused on the retailer-level, revealing that supermarkets are (i) growing
fast, (ii) becoming important, even dominant agents in the food supply chain and (iii)
implementing very different procurement systems than those observed in the traditional
food marketing system. In Kenya for example, essay 1 showed that supermarkets are
growing at an annual rate of 18%, had a 20% share of the urban food market in 2003 and
are rapidly developing the various pillars of the new procurement system (i.e.,
centralization, regionalization, preferred suppliers, specialized wholesalers, private
standards and grades). While indicating a likely differential impact of this more
demanding procurement system on farmer participation in the supermarket channel, the
analyses in this first wave of research on supermarkets in developing countries was for
the greater part limited in its scope to data from retailers and wholesalers. Based on
farmer surveys in the fresh fruits and vegetables (FFV) sub-sector, this essay aims to fill
this gap in the literature by analyzing the rise of supermarkets in Kenya from the farmer

perspective.

95

The strategic objective of this essay is to assess the effect of the development of
the supermarket sector in Kenya on F FV producers’ behavior and net incomes.

Controlling for product we have the following three research questions:

(1) What are the determinants of the farrners’ channel choice (supermarket channel

vs. the traditional channel)?

(2) What are the effects of participation in the supermarket channel on the farmer’s

production technology?

(3) What are the effects of participation in the supermarket channel on the farmer’s

' net income?

The research presented here focuses on kale16 growers, although some limited analyses
for tomatoes and bananas is included to get a broader perspective. These products were
chosen (i) because of their importance in terms of volume in Kenya’s produce sub-sector
and (ii) because they represent different risk-return trade-offs. Kale is a labor-intensive
crop (requiring intensive weeding) with a reliable yield (relative to agro-ecological
conditions), but a generally low market price, making it well-suited for risk-averse

smallholder farmers who have a high labor-to-capital ratio. Tomatoes are more capital

 

'6 Kale, a leafy green vegetable (member of the cabbage family) is the most widely grown and consumed
vegetable in Kenya. Its low price and high nutritional value make it especially important in the diet of the
poor who consume it with the local (maize-flour-based) staple ugali (hence its local name sukuma wiki,
Swahili for ‘push the week’). In 2002, overall kale production in Kenya reached 317,000MT with an
estimated rural market value of Ksh2 billion, i.e., 30% of the total volume and 17% of the total value of
vegetable production respectively (MoARD 2002).

96

intensive (requiring irrigation for water control, more chemicals to control diseases, more
expensive good quality seeds, and so on) and their yield is more variable, but prices are
generally higher, thus attracting farmers with more capital, who can afford to take greater
yield risks in return for higher potential returns. In banana production, access to planting
materials (e.g., tissue culture) is a critical issue as demand for these materials outstrips
supply (MoARD 2000), so that less mobile and less informed smallholder farmers have
less access to it.

The essay is structured as follows. The next section briefly lays out the conceptual
approach to the research questions. Section 3 indicates how supermarket-channel farmers
and traditional-channel farmers were defined in this study and describes the data that
were collected. Section 4 compares the characteristics, the production practices and the
marketing practices of farmers directly supplying supermarkets and farmers supplying the
traditional system of brokers and wholesalers. Section 5 uses the case of the kale supply
chain to assess the determinants of channel participation (sUpermarket channel vs.
traditional channel) and analyzes the production technology and net income effects of this

participation. Section 6 concludes and provides recommendations for development policy

and programs.
2. Conceptual Approach
The rise of supermarkets in Kenya is here modeled as a shock to a domestic F F V supply

system that has shown few dynamics over the last few decades, apart from a steady

grth in volumes and in the number of rural producers in response to urbanization

97

(Dijkstra 1999a). On the one hand, supermarkets may pay higher prices to farmers and/or
may offer them more demand reliability both in terms of price and volume. On the other
hand, meeting the supermarket’s higher standards for food safety and quality and other
delivery conditions may imply the presence of (or additional investments in) certain
production and marketing capital as well as an increased need for working capital to deal
for example with higher input use and buyer credit. In other words, the rise of
supermarkets in Kenya is hypothesized to contribute to an industrialization of the food
system as observed in other countries (Reardon and Barrett 2000). Here we shall take an
essentially micro-economic approach, focusing on how this shock impacts the behavior
of FFV producers as individual agents.

First, we model the farmer’s decision on whether or not to sell to supermarkets as
a standard static adoption decision, where the adoption of X (in this case the supermarket
channel) is determined by the incentives for and capacities of farmers (Peder et a1. 1985).
The different requirements of supermarkets are hypothesized to determine a K“ vector of
threshold capital requirements, embodying a technology, which suppliers must have
before the supermarket channel enters their opportunity set. Different technologies as
embodied in different capital vectors (e.g., a labor intensive K1 * and capital intensive
K2*) may exist, with both giving farmers the capacity to meet the supermarkets’
requirements. Furthermore, K“ may only be weakly enforced by supermarkets and will
likely be a moving target for farmers (supermarket requirements become more stringent
over time). K“ may include physical capital (e.g., land, vehicles), human capital (e.g.,
business experience), financial capital (e.g., cash reserves) and organizational capital

(e.g., group membership). Farmers with a capital vector K>K* are expected to enter the

98

supermarket channel if the incentives are there. Farmers with K<K* are excluded from
the supermarket channel. The latter may be undesirable from a social welfare point of
view, if the supermarket channel represents higher net incomes for farmers and if the
smallholder farmers who make up the bulk of Kenya’s population are being excluded
from it. Our conceptual model then takes on the following general reduced form (i.e., the
input demand function as derived from the farmer’s profit function; Sadoulet and De
J anvry, 1995):

channel choice = f(p, r, o, k, z),

where p, r, o, k and 2 represent output prices, input prices, risk factors, quasi-fixed capital
and shifters respectively. Output prices, input prices and risk factors (including
transaction costs) are in part implicit to the channel choice (which determines governance
structure) and are further determined by farm characteristics such as its location and size.
Therefore input and output prices and risk factors will not be directly entered in the
implementation model (section 5.1). Given the limited geographic spread of the sample
populations (see next section) and thus too limited variation across the sample for this
variable, location as a key shifter will also be excluded. This implies that our channel
choice model will mainly be determined by quasi-fixed capital. The latter both represents
resources internal to the farm and at the same time co-determines access to external
resources. For example, size of the land owned by the farmers is a direct resource as well
as an indicator of access to credit (Carter and Wiebe 1990).

Second, we assess the differential effect of channel choice on production

technology by comparing the production functions for farmers supplying supermarkets

99

and farmers supplying traditional marketing channel agents. The distribution of the
farmers over these two groups is not random but rather the outcome of a self-selection
process and this non-randomness may bias the estimated coefficients in the production
function. In order to control for this selectivity bias, we use Heckit’s two-stage method
(Heckman 1979). In the first stage, a probit model is used to estimate the selection model,
i.e., channel choice regressed on a set of exogenous determinants using the whole sample.
From this estimation, the inverse Mill’s ratio (1.) is calculated based on the residuals of
the selection equation. This ratio represents a summarizing measure which reflects the
effects of all the unmeasured characteristics which are related to channel choice (Smits
2003). In the second stage of the Heckman procedure, we estimate the production
function (the substantial model which is the real focus here) for both sub-populations
while adding the lambda variable as an explanatory variable to control for self-selection
bias.

Third, we assess the net income effect of channel chOice using gross margin
analysis. We then place this net income effect in the broader context of value chain
theory (Kaplinsky and Morris 2001). According to value chain theory, the following
dynamic models the growth process of a firm: (1) the performance of firm A depends on
the performance of A’s network partners and vice versa (systemic efficiency); (2) the
participation of A in a supply chain takes on a particular governance structure (system
access) depending on A’s performance, the policies of the government and the channel
captain of A’s supply chain and the location and network relationships of A; (3) the
benefits accruing to A depend on the governance structure for A, the performance of A

and A’s market power (distributional effects); (4) the amount of benefits accruing then

100

determine the degree of upgrading, which in turn affects the performance of A (coming
full circle: system dynamics). We will assess this model by comparing (income and risk
related) benefits accruing to farmers, governance structures and upgrading/growth in the

supermarket channel vs. the traditional channel.
3. Definitions and Data

Farmer Surveys

The central objective of this essay is to better understand the nature and behavior of the
FF V farmers who are supplying supermarkets in Kenya. There are basically three types
of FFV suppliers to supermarkets (essay 1): importers (14% of supplies), brokers or other
middlemen relying mostly on smallholder producers (43% of supplies) and farmers
supplying direct to the supermarket (43% of supplies). Of the latter, 23% come from
small farms (<10 acres), 42% come from medium-sized farms (IO-40 acres) and 35%
come from large farms/plantations (>40 acres). Farmers as direct suppliers of FFV can
further be classified into listed and unlisted suppliers, i.e., based on whether or not they
are part of the list of regular FFV suppliers to the supermarkets. In 2003, 125 out of an
estimated 400 direct F F V suppliers to the leading two chains were listed, with the non-
listed suppliers mainly consisting of smallholder farmers and brokers ad hoc addressing
shortages not yet resolved by listed suppliers (according to our interviews with the
supermarkets). In this context, the target population is here defined as ‘the farmers listed
as direct FFV suppliers to Uchumi and Nakumatt supermarkets’. This population was

selected for the following four reasons. First, within the next five years, 85-90% of the

101

locally sourced F F V sold in supermarkets are expected to be supplied directly by farmers
(or farmer groups), as opposed to through traditional brokers (in 2003, 50% was directly
supplied by farmers; essay 1). Second, as supermarkets develop their FFV procurement
system these direct supplies will come from listed suppliers only. Third, farmers
supplying supermarkets indirectly through traditional brokers feel no impact from
supermarkets because the brokers’ buying behavior is not different, as these are, again
according to our interviews with the supermarkets, the same brokers supplying wholesale
markets in Nairobi (i.e., they are not specialized brokers working with the same group of
farmers over time). Fourth, 90% of FFV sold through supermarkets are sold through the
two leading chains, Uchumi and Nakumatt (essay 1). We shall further refer to this
population of ‘farmers listed as direct FFV suppliers to Uchumi and Nakumatt’ as
‘supermarket-channel farmers’.

In order to assess to what extent these supermarket-channel farmers are different
from farmers supplying through the traditional F FV system, a Similar data set was also
collected on farmers who supply directly to traditional marketing agents (brokers,
wholesalers, open air markets), but not to supermarkets. Since this is a large and
heterogeneous population, we focused on farmers in selected key production regions for
selected produce items bound for the Nairobi market. The selected products are kale,
tomato and banana, which are in terms of volume the two most important vegetables and
the most important fruit in the domestic market. The production areas for these products
were selected based on their relative importance as indicated in interviews with operators
at Nairobi’s key FFV wholesale markets (Wakulima, Gikomba, Kawangare, Kangemi).

We shall further refer to this population as ‘traditional-channel farmers’.

102

Two farmer surveys were conducted. The first farmer survey took place in the
period September-November, 2003 and focused on the capacities of the farmers and the
marketing methods they used. A total of 115 farmers were interviewed, comprising two
sub-samples. The first sub-sample consists of 49 supermarket-channel farmers randomly
selected from the supplier lists provided by Uchumi and Mugoya, the (then) specialized
FFV wholesaler of Nakumatt (these 49 farmers equal 40% of the listed FFV suppliers).
One of the requirements for being a listed supermarket-channel farmer is that the farmer
can easily be reached by phone. Thanks to the emergence and rapid diffusion of cellular
phones in Kenya, many farmers, not just the rich ones, have (access to) a phone (e.g.,
30% of the traditional-channel farmers in our research could be reached by cellular
phone). This allowed us to contact the supermarket-channel farmers by phone to set up a
time and place to meet for the interview. If a farmer could not be reached or refused to
participate, he or she was replaced by the next farmer on the list (which consists almost
eXclusively of individual farmers). The interviews usually took place on the farm, but on
occasion at a supermarket after delivery or at an in-town office (as we shall discuss in
more detail in section 4.3, quite a few of the key supermarket-channel farms started out
as “hobby-farms” of those with well-paying formal jobs in Nairobi who turned
commercial in response to the rise of supermarkets). Most of these farmers produced a
wide variety of FF V items (most of them marketed to the supermarkets) but each farmer
was interviewed in detail only for her or his highest volume produce item”. In terms of

location, around 80% of the supermarket-channel farms were located in favorable F FV

 

'7 These items are: kale (12 growers), tomato (5 growers), banana (5 growers), spinach (12 growers),
papaya (3 growers), lettuce (2 growers), cabbage (2 growers), pineapple (2 growers), and broccoli,
mangoes, capsicum, black nightshade (mnavu), French beans and potatoes (1 grower each). In the
comparative part of our analysis we focused on the 5 tomato, 12 kale and 5 banana growers in this sample.

103

zones within a mom radius around Nairobi (mostly in the highlands north and north-
west of the capital), the same zones where most of the traditional-channel farmers are
located. The other 20% were located either farther from Nairobi or in the less fertile rural
areas south-east of Nairobi (which have the advantage of lower prices and greater
expansion possibilities due to the far lower population density). The average distance
from the farm to Nairobi is 72km (i.e., closer than the traditional channel farmers in this
study). For this sample we found that 25% of them are small farms (<10 acres), 43% are
medium-sized farms (between 10 and 40 acres), 18% are large farms (between 40 and
200 acres) and 14% are plantations (>200 acre). This closely matches the distribution of
the FFV supplies over different sizes as given by the supermarkets.

The second sub-sample consists of 66 traditional-channel farmers (22 kale
growers, 22 tomato growers, 22 banana growers), all of them located in zones which are
well-suited for these crops and which are key production zones for the Nairobi-bound
supplies of these produce items”. In terms of distance, these farms are located farther
away from Nairobi than the supermarket-channel farmers (less than 70% are within a
mom radius around Nairobi, while the average distance to Nairobi is 122km). With no
crop specific population frameworks available, farmers were selected (within a division)
based on a judgment sample with the assistance of local traders and/or the division’s
horticultural development officer. The latter were asked to select a sample that was in
terms of farm size representative for the division. The average farm size for the
traditional-channel farms in our sample is 5.2 acres, which is similar to the average farm

size indicated in other reports as varying from 3 to 7 acres depending on the specific

 

“The selected districts were Kirinyaga (22 farmers) for tomatoes, Kiambu (l2 farmers), Thika (6 farmers)
and Nyandarua (4 farmers) for kale and Meru ( l4 farmers) and Muranga (8 farmers) for bananas.

104

location (e.g., Dijkstra 1992, Ministry of Agriculture 2003, 2004). Of the traditional-
channel farmers sampled in this survey, 85% were smallholders and 15% were medium-
sized.

The second farmer survey (August 2004) focused on technology choices and net
income effects. For this survey we focused on kale only. Kale was selected because it is a
produce item for which there is a sufficiently large group of suppliers to supermarkets
and for which supermarkets have developed their supply system the most with nearly all
of the kale being supplied directly to the supermarket by listed farmers (and thus
providing an interesting case-study of the farm-level impact of supermarkets). A total of
51 farmers were interviewed in this survey, again comprising of two sub-samples. In the
first sub-sample, we selected 14 out of 30 listed farmers supplying kale to Uchumi and/or
Nakumatt (this sample thus represents roughly half of the kale suppliers to these chains).
Of these farms, about 60% were located in the same key production area from where we
selected the traditional-channel sample (i.e., Kiambu district, well-suited for growing kale
in terms of agro-ecological conditions and, because of its proximity to Nairobi, well-
suited for marketing the highly perishable kale). The other 40% were located in the less
fertile (peri-urban) zones south-east of Nairobi, where as mentioned above, land is more
available and thus less expensive. The average (total) farm size in this sub-sample is 31
acres and the average distance to Nairobi is 36km (i.e., ten times the size of the
traditional-channel farmers described in the next paragraph and located slightly closer to
the capital).

To have a comparable sample (second sub-sample), we then interviewed 37

traditional-channel farmers located in two divisions which were identified as key kale

105

production areas for the Nairobi market (Lari and Limuru divisions in Kiambu district).
Kale growers in these two divisions represent a production volume that is 50% of the
estimated consumption in Nairobi. The average distance to the market is, as in the first
farmer survey, greater for traditional-channel farmers than for supermarket-channel
farmers (49km vs. 36km). As in the 2003 survey, farmers were selected based on a
judgment sample with the assistance of divisional horticultural development officers. The
average farm size of the traditional-channel farmers in this sample is 3 acres, which is in
line with the officially reported average farm size of 3 acres (Ministry of Agriculture

2004).

Semi-Structured Interviews

In addition to the two farmer surveys, additional semi-structured interviews were
conducted to get a better understanding of various aspects of the domestic FFV system in
Kenya. These included interviews with: (1) selected input suppliers to obtain cost data on
certain inputs; (2) key experts on the different technology options; and (3) the successive

intermediaries downstream from farmer to retailer in the traditional FFV market channel.
4. The Nature of Supermarket-Channel Farmers
4.1 Supermarket-Channel Farmers Compared to Traditional-Channel Farmers

In this section we will compare supermarket-channel farmers and traditional-channel

farmers with respect to a set of key characteristics for the three selected products: tomato,

106

kale and banana. The data indicate that there are substantial differences between
supermarket-channel and traditional-channel farmers in terms of landholdings, labor use,
gender, human capital, physical capital, diversification, organization capital and
marketing practices. In this section we shall discuss only the findings where the

significance of the difference is at the 10% level or higher.

Land Related Diflerences

Supermarket-channel farms are on average much larger, in overall farm size, than
traditional farms (Table 13). This holds across the three products, but is especially stark
for the kale farmers where the ratio between the two sizes is 12 to 1. One reason for this
may be that kale is a less risky crop than tomatoes (requiring fewer inputs, less capital)
and therefore involves more smallholder farmers in the traditional channel thus reducing
the average overall farm size in this group. Not only is the overall farm size, but also the
land devoted to the specific crop (tomato, kale or banana) is greater for the supermarket-
channel farmers. The main reason is that given their higher land availability supermarket-
channel farmers will optimize by bringing an area under cultivation that is in line with the
size of the orders they get from the supermarkets, taking into account that for a year-
round delivery (as preferred by the supermarkets) several plots of the produce item in
various successive stages of the harvest cycle are required. By contrast, the traditional-
channel farmers use part of the farm size for food-crop production (beans, maize), for
dairy, for their home-stead, and so on, leaving only small plots (relative to the

supermarket-channel farmers) for FFV production, even if these plots represent a large

107

portion of the farm size (e.g., for kale on average nearly 25% of the overall farm-size,
compared to only 7% for supermarket-channel farmers).

Other patterns, related to technology and land-use, which hold across the three
selected products are that supermarket-channel farmers have less of their land under
cultivation (e.g., in the case of kale 71% versus 87% for traditional-channel farmers), but
have a larger percentage of it under irrigation. Both patterns are related to the need to
supply supermarkets on a regular basis throughout the year. Fallow periods, as a part of
crop rotation cycles, are facilitated when farms are large and enough land is available to
take the place of the land under fallow. Another reason for the higher percentage of the
land not farmed is that supermarket-channel farmers have bought land for future
expansion as they are growing fast (see section 5.5). Irrigation is a condition sine qua non
for regular year-round supply and the most critical requirement demanded by
supermarkets (essay 1). Again, the difference is most pronounced for the kale farmers
(75% of the farm size is irrigated for supermarket-channel fMers versus only 18% for
traditional-channel farmers). As before, this results from the large number of small farms
which have specialized in kale and which do not have the capital to invest in irrigation
systems.

With regard to landownership, clear distinctions emerge across the products.
Amongst kale growers, supermarket-channel farmers basically own all of their land while
only two thirds of the farm land of the traditional-channel farmers is owned by them. This
is mainly because of land provided for free by parents to their children in the selected
divisions for kale (relatively densely populated areas with limited opportunities to buy or

rent land). Supermarket-channel farmers who supply tomatoes on the other hand opt

108

mostly for leased land relative to their traditional competitors. One reason is that the
tomato growing regions selected in this study have a relatively dense network of rivers so
that there is no need to install boreholes and neither are other permanent structures (e. g.,
greenhouses) used for tomato growing. The presence of such fixed capital assets would
create an incentive to buy rather then lease land in order to secure them. For banana
growers there is no significant difference in landownership (as a percentage of the farm
size) between supermarket-channel farmers and traditional-channel farmers. The
observation that many coffee growers shifted to banana growing after sustained periods
of depressed coffee prices (MoARD 2002) and the more long-term nature of the
cultivation of bananas are two likely determinants of relatively high degrees of land-
ownership for both groups of farmers (nature of crop provides incentive to secure
investment through land-ownership). Specialization in horticulture (as a percentage of the
farmed land) appears to be slightly higher for supermarket-channel farmers, except for
kale. We further found a negative correlation between land ownership and specialization
in horticulture (significant at the 1% level), indicating that leased land is used more
intensively for F FV. This may be the result of leased land being leased for a particular
purpose (e.g., to grow only tomatoes) and for a particular time period, potentially short
enough not to care too much about depleting the soil because of intensive cultivation of
one particular crop (another, more fertile piece of land may be leased for the next cycle).
It can further be expected that a resource (like leased land) with a specific use limitation
(duration of the lease) and explicit cost (rent payment) provides more of an incentive to

get the highest possible return from it.

109

Table 13: Land Related Differences (supermarket- vs. traditional-channel farmers)

 

 

Crop Tomato 1(a_le Banana

Trad. Super. Trad. Super. Trad. Super.
Farm characteristic Farm. Farm. Farm. Farm. Farm. Farm.

m=22) (N=5) (N=22) (N=12) (N=22) (N=5)
farm size (acres) 6.1 " 23.0 * 3.8 "' 45.6 "‘ 5.8 34.0
land owned (% of farm size) 60 "' 20 "' 66 **"‘ 99 *” 87 80
land irrigated (% of farm size) 93 "”" 100 " l8 *" 75 ""' 41 66
land farmed (% of farm size) 89 80 87 "' 71 "' 93 88
land used for horticulture (% of 90 *” 100 ‘" 83 65 52 66
land farmed)
Land used for crop (acres) 2.5 4.5 0.9 "‘ 3.4 * 1.5 * 9.3 "

 

Notes: ‘=significant at the 10% level, ** = significant at the 5% level, "* significant at the 1% level.
Source: author’s farmer survey 2003.

Labor, Gender and Human Capital Diflkrences

As expected given the difference in farm size, supermarket-channel farmers have more
permanent employees and more casual workers than traditional-channel farmers (Table
14). The average labor-to-land ratio (number of permanent farm workers per acre of
farmed land for the whole farm, not just the specific crop) is lower for supermarket-
channel farmers than for traditional-channel farmers, mainly due to a heavy reliance on
abundantly available family labor amongst the latter. For example in the case of kale,
79% of the permanent farm workers on traditional-channel farms are family members,
while for supermarket-channel farms 79% of the permanent farm workers are hired
employees. This pattern also holds at the aggregated level (i.e., sum of all farm workers
over sum of all farm sizes for each product-farmer type combination), although the
difference become less stark because the larger traditional-channel farmers are less labor
intensive. The aggregated numbers imply that the size of the farm is taken into account
and we found that labor-to-land ratio and size (measured as land farmed for all products)
are negatively correlated for traditional-channel farms (significant at the 1% level). For

example, kale farms (in the traditional channel) below five acres have a labor-to-land

110

ratio of 1.9, while farms above five acres have a labor-to-land ratio of 0.9. The high
number of hired permanent and casual workers on farms in the supermarket channel
implies that supermarket-channel farmers are (in absolute numbers) important providers
of job opportunities for rural households with little or no land. For permanent employees,
this also holds in relative (labor per acre) terms. In the case of kale growers for example,
the average farm-level number of permanent hired workers per acre of farmed land
(across all products, not just kale) is 0.9 for the supermarket-channel farmers and only 0.5
for the traditional-channel farmers. However, the opposite holds for casual workers: for
kale growers, for example, the average number of casual workers per acre of farmed land
is 2.1 for the supermarket-channel farmers and 1.3 for the traditional-channel farmers.
However, the comparison is less straight forward as the actual labor time for which
casual workers are hired is difficult to estimate (it was not estimated in this study which

merely asked farmers to indicate the number of casuals hired during peak times).

Table 14: Labor, Human Capital and Gender Differences (Supermarket vs.
Traditional-channel farmers)

 

 

Crop Tomato £31; Banana
Trad. Super. Trad. Super. Trad. Super.
Farm characteristic Farm. Farm. Farm. Farm. Farm. Farm.
(N=22) (N=5) (N=22) (N=12) (N=22) (N=5)
permanent employees(#/farm) 4 19 3 " 33 " 3 8
casual labor (#/farm) 10 " 17 " 4 " 19 " 4 *" 11 ”"
family labor (% ofperrnanent empl.) 47 *" 3 "" 79 "‘ 21 "" 74 ‘" 11 ”"
labor-to-land ratio (farm level) 1.4 0.9 1.6 1.1 1.1 0.5
labor-to-land ratio (aggregate) 0.9 0.9 1.1 0.9 0.6 0.4
schooling farm head (yrs) 9 " 13 " 7 "" 13 ”" 7 *" 15 ""
age farm head (yrs) 35 38 40 47 53 46
years in farming farm head 10.6 9.4 12.9 13.4 21.2 " 9.6 "
female labor (% of empl., farm level) 24 41 54 ” 39 " 47 20
female labor (% of empl,, aggregate) 22 59 51 34 36 10
female headed farms (%) 5 20 32 17 27 20

 

Notes: ‘=significant at the 10% level, " = significant at the 5% level, ""' significant at the 1% level.

Source: author’s farmer survey 2003.

111

The age of farmers does not appear to be different between the two groups of farmers,
and except for banana growers (some of whom have been in farming since the 19308 and
1940s), the same can be said about the years of farming experience the farmers have.
There is however a very clear pattern regarding the education of the farmers: traditional-
channel farmers on average have a primary education, while supermarket-channel
farmers on average have a secondary education. This can be explained mostly by the
observation that many of the supermarket-channel farmers had (or still have) formal jobs
in the private or public sector and started out as hobby-farmers (see section 4.2 for more
detail). Hence higher education gave access to formal, well-paying jobs, from there to
savings invested in farms and from there to supermarket channel participation.

The survey also allows us to make some observations regarding farm-level
gender-related aspects of the impact of the rise of supermarkets in the domestic FF V
supply channel. First, for kale and banana there appear to be fewer farms headed by
Women amongst the supermarket-channel farmers while the opposite appears to hold for
tomato, although no statistical differences could be found. Second, the average
percentage of full-time employees who are women is significantly lower for kale
supermarket-channel farmers than for kale farmers in the traditional channel. Within the
limitations of the data set, this pattern does not hold across all produce items as tomato
supermarket-channel farmers appear to hire more women than their traditional

competitors do.

112

Physical Capital Differences

Table 15 paints a stark picture regarding the diffusion of physical capital in the two
farmer groups. When we look at the kale farmers, we see that all of the supermarket-
channel farmers have a phone and their own (motorized) means of transportation, and
that a large percentage have an advanced irrigation system (i.e., drip or sprinkler), a
packing shed and electricity at the farm (either through a generator or from the grid). By
contrast, significantly less of the traditional-channel farmers use these technologies, with
(cellular) phones having reached the highest degree of diffusion amongst traditional-
channel farmers (3 out of 10 traditional-channel farmers have (access to) a cellular
phone). Other technologies such as greenhouses, shadow netting and cold rooms were
encountered, but only at very few farms, all of which are supermarket-channel farmers.
The latter is related more to the nature of these farms than that this has resulted from the
requirements of the supermarkets. For example, some of the supermarket-channel farmers
(about 10%) also supply (different produce items) for the export market and have set up
cold chain technologies for this. This implies that these farmers have an advantage if and
when domestic supermarkets would start requiring cold chain technology from their
produce suppliers.

While the small numbers of supermarket-channel farmers for tomato and banana
does not allow meaningful chi-square tests, it appears this wide divide in technology use
between supermarket-channel farmers and traditional-channel farmers also holds for
these two produce items. The relatively high percentages of traditional-channel farmers

using irrigation for tomatoes and bananas results from tomatoes needing at least furrow

113

irrigation and one of the selected banana growing areas having a government installed
irrigation infrastructure (main line).

The high rate of technology diffusion amongst the supermarket-channel farmers
(close to 100%) indicates that there probably is a threshold level capital vector (K‘)
which farmers must have in order to access the supermarket channel. Supermarkets
indicated to us that they select their F F V suppliers on the basis of their ability to become
reliable suppliers over the long term. This means that these farms must demonstrate their
ability to produce year-round (presence of irrigation system and year round source of
clean water) and to deliver with short lead times (presence of phone, transportation,
packing shed). The presence of these technologies is at present a very strong but not yet
strict requirement (not all these technologies have a diffusion rate of 100% for all
produce items), mainly because the FFV procurement system of the supermarkets is still
in an early, formative stage. A strict implementation of these physical capital

requirements is likely for the near future.

Table 15: Physical Capital Differences (Supermarket vs. Traditional-channel

 

 

farmers)
Crop Tomato .KLIS Banana
Trad. Super. Trad. Super. Trad. Super.
Farm characteristic Farm. Farm. Farm. Farm. Farm. Farm.
(N=22) (N=5) (N=22) (N=12) (N=22) (N=5)
farms with phone (%) 29 100 31 "" 100 "" 24 100
farms with irrigation system (%) 100 100 27 "" 92 "'" 55 100
farms with drip/overhead irrig. (%) 5 40 18 "”" 92 *" 32 40
farms with transp. vehicles (%) 14 80 9 "" 100 "'" 5 80
farms with a packing shed (%) 0 80 0 *" 75 *" 0 40
farms with electricity (%) O 20 5 *" 83 "* 5 20

 

Notes: "" significant at the 1% level.
Source: author’s farmer survey 2003.

114

Diflerences in Diversification, Organizational Capital and Marketing Practices

Table 16 compares supermarket-channel farmers and traditional-channel farmers over a
broad set of farm characteristics. In terms of diversification, two observations can be
made. First, we found no statistically significant difference in the percentage of overall
income coming from farming. Across the three products and two channels, farming is, on
average, the main source of income. Second, farmers growing tomatoes and kale for
supermarkets grow more than twice as many different horticultural crops and are
therefore far less dependent on the production and market risks of any particular crop.
This is in part the result of the greater availability of land for supermarket-channel

farmers.

Table 16: Differences in Diversification, Organizational Capital and Marketing
Practices (Supermarket vs. Traditional-channel farmers)

 

 

Crop Tomato _Ka_le Banana
Trad. Super. Trad. Super. Trad. Super.
Farm characteristic Farm. Farm. Farm. Farm. Farm. Farm.
(N =22) (N=5) (N=22) (N=12) (N=22) (N=5)
% of hh income fi‘om farming 87 82 81 60 87 60
# of different horticultural 2 " 5 * 3 ”" 7 "“ 3 2
products grown (at a given time)
farms collaborating with other 46 0 32 33 29 80
farms (%)
farms member of coop/assoc. (%) 18 0 73 "" 8 "" 77 0
farms with livestock (%) 41 0 68 92 96 60
output marketed ofcrop (%) 84 "'" 99 "" 72 ""' 96 "W 74 ”* 100 ’”
farms keeping farm records (%) 64 80 33 "" 92 *" 18 100
farms transporting to buyer (%) 23 100 9 "" 100 "" 14 100
farms getting phone-orders (%) 0 80 5 ”" 100 "" 0 100

 

Notes: ‘=significant at the 10% level, " = significant at the 5% level, "‘ significant at the 1% level.
Transportation to buyer is for transportation beyond the main road. Most kale farmers will take produce
to brokers near main road on a donkey cart.

Source: author’s farmer survey 2003.

In terms of organizational capital, table 16 shows some interesting variation. While the

degree of collaboration amongst kale farmers is not related to the channel they operate in,

115

the nature of this collaboration is related. Kale farmers in the traditional channel
collaborate mostly on the marketing side, bulking their produce to facilitate transporting
it themselves to the wholesale market where they get a better price (this type of
collaboration is more prevalent as the farms are located closer to Nairobi). Kale farmers
in the supermarket channel, who know each other in part because of the farmer meetings
organized several times per year by the supermarkets, mostly collaborate on production
issues (sharing of best practices). For tomatoes, it is the traditional-channel farmers who
collaborate more because it allows them to overcome their small size and jointly transport
their produce to the wholesalers in Nairobi. This is related to the nature of the tomatoes
which require proper synchronizing of production and marketing (once at the right
ripening stage for the intended market, tomatoes need to be harvested relatively quickly).
This means that tomato farmers in the traditional channel (who for the greater part have
no transportation) are more at the mercy of (collecting) brokers, a dependency they try to
avoid by taking the marketing of their tomatoes in their own hands through collective
action. Tomato growers in the supermarket channel on the other hand have their own
transportation (as well as a larger volume) and therefore have less incentive to collaborate
with other farmers. For bananas on the other hand it is the supermarket-channel farmers
who collaborate more. This is because these farmers (1) want to collaborate on the
production side, trying new banana varieties and new propagation techniques such as
tissue culture bananas; and (2) these farmers are trying to develop new marketing supply
chains for bananas, with some farmers taking the lead and organizing other growers.
While point (1) above results from the fact that supermarket-channel farmers are better

educated and thus have better access to new technologies (not from any specific initiative

116

of the supermarkets), point (2) is to some extent stimulated by the growth of
supermarkets. For example, one of the supermarket banana suppliers (who uses tissue
culture bananas) found that the Kampala banana was more popular amongst consumers
than the Cavendish banana he was growing, but realized that the Cavendish’s longer shelf
life would still make it the winner in supermarkets. Having been a supplier to Uchumi
since 2001, he has worked out a concept for a supplier organization (with collection
points for multiple farmers, cold storage, and ripening rooms in Nairobi) that could
provide the supermarkets with the banana supply they need (year-round supply of good
quality bananas with a long shelf-life). The main (basically only) obstacle to the
realization of his concept is a lack of capital.

Table 16 indicates that there is greater participation in cooperatives or
associations amongst traditional channel farmers than amongst supermarket-channel
farmers. An important reason for this is that the former are linked in cooperatives or
associations related to non-FF V products. For example, an important diversification
strategy of traditional-channel farmers is to go into dairy and many of those with dairy
cows are a member of a dairy cooperative (this is especially the case for the kale growers
in Kiambu district)”. While about half of the supermarket-channel farmers also have
livestock (especially kale farmers), very few (less than 5%) of them are members of a
dairy cooperative. Other traditional-channel farmers are part of formal self-help groups
(especially women), mostly with the objective to pool financial resources and create a
revolving fund from which members can draw to make investments or deal with

emergencies. Some of the traditional tomato and banana farmers in Kirinyaga and

 

‘9 Livestock/dairy provides a perfect complement to FF V production as it transforms the nearly worthless
FF V wastage into meat, milk and manure. For smallholder producers dairy is also important because it
creates a more continuous income to offset the more erratic income from F FV.

117

Muranga districts also grow coffee and are members of coffee coops. Only two F F V
cooperatives were encountered within the study’s sample, both for traditional-channel
farmers in Muranga district. The first is the Mukago Bio-Banana Group which is focused
on organic growing methods for bananas intended for the domestic market and has 60
members who are organized around a farmer field school. The other is the Koome Self
Help Group which has 156 members and is focused on the marketing of horticultural
crops. Farmers did indicate that coop membership gives them greater access to low-
interest loans which they could use in their FFV farming and which could facilitate
market access. However, the data presented here clearly indicate that coop membership is
not a determinant of access to the supermarket channel at this point in time. The main
reason for group formation amongst FF V farmers is to overcome a small scale which
stands in the way of efficient marketing. Supermarket-channel farmers, because of their
size, have less need to group themselves for marketing purposes.

Finally, table 16 points to some differences in the marketing practices of the two
farmer groups. First, supermarket-channel farmers market a higher percentage of their
crop than do traditional-channel farmers. This is in part related to the latter’s smaller size,
which implies that home-consumption plays a more significant role. However, FFV are
mainly intended for selling to the market as for all three produce items and for both types
of farmers more than 70% of the production is marketed (it is nearly 100% for the
supermarket-channel farmers). Second, the way in which the produce is marketed differs
between the two supply channels in important ways. Supermarket-channel farmers.
receive their orders by phone, transport the ordered produce to the supermarket and keep

records of their deliveries in order to be able to follow up with payments. By contrast,

118

traditional-channel farmers just harvest when they think the market price is good (or
when they can no longer postpone harvesting), sell the produce either from the farm or
take it by bicycle or donkey cart to a collection market near the main road and keep no
records of their transactions. These differences hold across the three produce items as
could be expected given that they are directly determined by channel choice (by the

supermarkets).

Differences in Marketing Capacity

Looking across the four tables in this section, it becomes apparent that supermarket-
channel farmers differ greatly from traditional-channel farmers in their marketing
practices because they differ greatly with respect to their marketing capacity. The latter
has four components. First, there is a physical capital component. Nearly all supermarket-
channel farmers have a pick-up truck, a packing shed and a cell-phone. Renting
transportation provides less control and farmers may supply markets too late (loose a
buyer or face lower prices). Second, there is a human capital component. Supermarket-
channel farmers have a larger labor force (needed to allow for short order-cycles) and
better business management skills. The latter follows from their higher level of education
and more extensive business experience (e.g., having had their own non-farm business or
experience with supplying FFV to exporters, institutions, wholesalers). Third, there is a
risk component. Supennarket—channel farmers’ risk management strategies are more
market-focused: more diversified income, more market options, more produce items.
Compare this to small-scale farmers whose risk management tools focus on the

production side and include keeping quasi-fixed capital low, minimizing cash outflows

119

and integrating horticulture with livestock/dairy. Fourth, there is a working capital
component. Accessing the supermarket channel greatly increase working capital needs, as
payment periods shift fi'om zero days (the cash-on-delivery used in the traditional
channel) to periods of up to 30 days”. Again, these differences over a broad set of
capacity variables indicate the existence of a threshold capital vector at the entrance of
the supermarket supply channel.

Given the limited marketing options available to smallholder farmers, it is not
surprising that our survey reveals that 60% of the traditional-channel farmers say they
have never thought about supplying supermarkets. The 40% who did think about it either
had no idea on how to initiate the contact or saw another obstruction that closed off the

channel for them (e. g., no transportation).
4.2 The Relative Impact of Export Channel Farmers in the Supermarket Channel

Exporters and export-channel farmers are only of limited importance as suppliers of F F V
to supermarkets, with only 10-15% of supermarket-channel farmers being involved in
export markets as well. Exporters are critical in some produce lines (mostly the higher
value, lower volume items such as French beans or avocado), and absent in other produce
lines (most domestic market fruits and vegetables, e. g. banana, tomato, kale). Mainly this
is because there is little overlap in product types. For example, for kale, tomato and

banana (which combined represent in volume terms 50% of the domestic FFV

 

2° The up and downs of supermarkets can increase the working capital challenge. Cash flow difficulties at
Uchumi in 2004 led to partial payments to FFV suppliers. One farmer for example indicated that the
supermarket held two months worth of his turnover in its accounts payable to him. Only farmers with
sufficient financial working capital can survive such cash-flow pressure and remain in the supermarket
channel.

120

production), exports are insignificant (less than 100MT in 2002; MoARD 2002). These
produce items find ready markets in Kenya and F FV exporters have instead focused on
(directed their capital toward) higher value produce items such as French beans, snow
peas, Asian vegetables (okra, eggplant) and avocadoes (these exports also include flesh-
cut, shelf-ready (and at times even labeled) vegetable packs for supermarket chains in the
EU). For fruits, the only two items which are of importance to both the export and the
domestic market are mango and passion fruit, but even then only 4% of production is
being exported.

The importance of exporters thus critically depends on the type of produce. The
three main instances where the export and the domestic supermarket channels do overlap
are: (l) semi-processed flesh vegetable packs of typical export produce items such as
French beans or snowpeas (90% comes flom exporters); (2) Asian vegetables, such as
okra and eggplant, mainly directed to the large Asian community (50% comes flom
exporters); (3) fluits which are of some importance for exports, i.e., pineapple”, avocado,
mango and passion fluit (20-100% coming from exporters, e.g., 100% for pineapple, 20%
for mango).

Farmers who are linked to both the export channel and the domestic market
channel (like the 10% of the supermarket-channel farmers mentioned here), in many
cases do so for different produce items (or even non-produce items: two of the
supermarket-channel farmers in our sample exported flowers). For example, a farm may
produce French beans as an outgrower for one of Kenya’s major FFV exporters and at the

same time produce tomatoes as a listed supplier to supermarkets in the domestic market.

 

2' Fresh pineapples are not very important in Kenya’s FF V exports, but Kenya’s largest pineapple producer
(DelMonte) is a major exporter of canned pineapple, while at the same time it sells flesh pineapples and
pineapple juice in the domestic market (including supermarkets).

121

Typically these farmers are: (1) first involved in the export channel for a particular crop
(say French beans); (2) using the experience and earnings from exports to expand their
capital vector (land expansion, additional vehicles, construction of a packing shed,
familiarity with good agricultural practices, and so on); and (3) then using their increased
capital assets to also enter the supermarket channel as producers of domestic market crop
(say spinach). The other instance where the export channel and the supermarket channel
overlap is when the leading exporters divert some of their production (second grade
unless there is oversupply) flom the export market to the domestic market (see for
example the case of Sunripe, section 4.4). And even here, there is a tendency to move
away flom the exporter: Fresh ‘11 Juici, Nakumatt’s specialized FFV supplier, is doing
most of the flesh-cut processing in-house (in part because supermarkets want affordable,
first grade produce for their customers). However, the volumes involved are of relatively
minor importance to both the exporter (almost all is exported) and the domestic
Supermarket (almost all F FV sales are for domestic market crops; although this is likely
to increase in importance as supermarkets increase their market share in the FFV market).

For farmers, the difference in accessibility between the supermarket channel and
the export market channel depends in great part on their size. Whereas (some)
smallholder producers can access the export channel through various outgrower
schemeszz, this is far less feasible in the supermarket channel since few such outgrower
schemes exist and hence fewer smallholder producers are involved (apart flom through

traditional brokers, but supermarkets are moving away from brokers; essay 1).

 

22 Albeit increasingly less so due to the increasingly stringent food safety standards in the export channel
(see for example Dolan and Humprey 2000).

122

4.3 Supermarket-Channel Farmers Categorized

When we look at the supermarket-channel farmers as a group, it becomes apparent that to
a large extent they represent a newly emerging class of farmers in Kenya. In the previous
two sections we made the case that supermarket-channel farmers are distinct flom both
the typical traditional-channel farmers and the export-channel farmers. So where then
does this emerging class of farmers come flom? Based on our random sample of 49
farmers flom the leading superrnarkets’ FFV supplier lists, we can distinguish four types
of supermarket-channel farmers: (1) specialized farms, whose main focus is the
production of FF V for domestic supermarkets; (2) intensifying farms, whose main focus
is not FF V for the domestic market, but who use part of their resources to also supply
FFV to supermarkets; (3) specialty product farms, whose main focus is value-added FF V
(e. g., vegetable packs) which they supply to supermarket amongst others; and (4) market
diversifying farms, whose main focus is FFV for the domestic market, but who do not

have the domestic supermarkets as their main focus.

Type 1: Specialized Farms

The first type consists of recently established farms (in the last 10 years) that linked up
with supermarkets in the last 5 years, started shifting to supplying mainly or even only
domestic supermarkets and experienced fast grth since. An estimated 43% of the
farmers fall in this category, all of them mainly growing vegetables. In many cases, the
farmer is a well-educated individual who has or had a primary job in Nairobi (e.g.,

private or public sector employees, small business owners) and invested his or her

123

savings in land (landownership also has an important socio-cultural meaning in Kenyan
society). This land was at first either not used, used for non-intensive farming (e.g.,
maize) or used for intensive farming (horticulture), but remained mostly a hobby for its
owner. Some of the more entrepreneurial farmers in. this group took samples of their
produce to the supermarket’s produce procurement manager in order to get a trial order
and farmers successful in getting the quantity, quality and delivery right for the initial
supplies succeeded in linking up with the supermarkets in a more long term relationship
(i.e., the usual way farmers got linked up with supermarkets once the latter started to
really expand the FFV category). Once they gained entry, these farmers quickly realized
the potential (increasing supply orders flom supermarkets coming in), but found that
continued access would require them to make additional investments in production and
marketing technology and run their farms under more professional management. In many
cases, it implied an exit from other professional engagements (albeit not necessarily for
both spouses in a household). In other cases, farms were set up flom the start for full-
time, commercial farming.

A typical example of a farmer of this type is L F arm23 . L farm was started in 1999
by a former employee of the Central Bank of Kenya who over time had build up a three-
branch supermarket chain in the greater Nairobi metro area. L-farm initially supplied its
own supermarkets. Since its start in 1999, the farm has specialized in leafy green
vegetables (because of their high turnover) and almost exclusively supplied supermarkets
(because they offered the best prices and presented the lowest transaction costs as they

buy in large volumes). As the supermarket sector was growing and becoming more

 

23 Farm names are withheld throughout this essay in order to comply with the MSU/UCRIHS stipulations.

124

competitive, the farmer felt his competitive advantage was in farming and so he sold his
supermarkets for premium prices to the leading supermarket chains and invested the
proceeds in the expansion and upgrading of his farm. The farm is located 35km south of
Nairobi in the dry, very sparsely populated areas of Kiserian district on three pieces of
land bought flom the pastoral Masai tribe. It is relatively inexpensive land with almost
boundless expansion options, only limited by the presence of aquifer layers. The latter
must be accessed by boreholes of which L Farm has installed two. Given the limited
availability of water, 64% of the farmed area is under an efficient drip irrigation system
which allows the farm to grow and supply F F V year-round. Supplies of leafy vegetables
(e.g., kale) to the Nairobi market show strong seasonal patterns (as most of the
production takes place under rainfed conditions) with shortages (especially of good
quality produce) occurring during several months each year. L farm’s ability to produce
good quality leafy vegetables in those low-supply months thus addresses the
supermarket’s key requirement of year-round supply ability of farmers.

L Farm started flom an initial size of 30 acres in 1999. In 2001 it began supplying
the leading supermarket chains, Uchumi and Nakumatt, growing along with them as they
expanded their produce sections. By 2004, L Farm had grown to 110 acres (of which 55
are farmed, the rest set aside for future expansion), had four transportation vehicles and
three hand-tractors and employed 50 workers. Not withstanding this high growth rate, the
farmer indicated that he could have increased his production much faster if affordable
credit would have been available to invest in additional boreholes. The absence of such
credit is the farmer’s most critical problem and one he has been trying to discuss with

supermarkets, requesting them (1) to provide farmers like him (who have all the right

125

capacities: market knowledge, size, irrigation system, close to Nairobi, and so on) with
formal supply agreements that can be used to demonstrate credit worthiness of farmers
and (2) to even broker with commercial banks or other lending institutions to provide
affordable credit to their (listed) FFV suppliers. In conclusion, the case of L Farm
demonstrates the emergence of a list of preferred FFV suppliers in the supermarket’s
strategy to assure supply consistency. This is similar to what has been observed

elsewhere (see for example Berdegue et all. (2004) for the case of Central America).

Type 2: Intensijying Farms
The second type of supermarket-channel farmer consists of larger farms with extensive
business experience whose main economic activity is not FF V production for the
domestic market, but who use part of their land for the latter in order not to leave it laying
idle. 'Ihese main activities (which are to these farms more profitable than the supply of
FFV to supermarkets) are varied and include flowers or vegetables grown for export, tea,
livestock and dairy processing, amongst others. In most cases, F F V were first supplied to
traditional-channel agents, such as wholesalers, and only later to domestic supermarkets
(when these became bigger sellers of FFV). We can classify 22% of the supermarket-
channel farmers in this category.

A good example here is SR Farm. SR Farm is an exporter of flowers to the EU.
The 250-acre farm has 258 employees and was a coffee plantation until 2000. With
coffee prices depressed, the farm shifted to alternative, more lucrative horticultural
products with 60 acres under horticulture by 2003 (the remaining 190 acres are still under

coffee but are not harvested (because of the low prices) nor uprooted because of

126

government restrictions and for strategic reasons, i.e., a future increase in coffee prices).
Of these 60 acres, 12.5 acres were allocated to the production of roses for export which
are now the main focus of the farm (they are the most profitable part, high turnover per
acre). Another part of those 60 acres is devoted to vegetables for the export market (e. g.,
8 acres of French beans). However, most of the 60 acres under horticulture is used for the
production of FFV intended for the domestic market For the latter they produce a wide
variety of items, including 3 acres of bananas, 4 acres of cabbages and 14 acres of kale.
Of the kale about 60% was marketed to the supermarkets and 40% to traditional local
markets. Their expansive, advanced irrigation system (covering 100 acres) allows them to
produce F F V during the off-season. The farm is capital intensive using 5 tractors, two
trucks, charcoal and electric coolers, modern packing sheds and greenhouses (1/2 ha).

For supermarkets, the main attraction points of SR Farm are its large supply
capacity, its ability to supply constantly throughout the year and its business skills, which
make it easy to order flom them (e.g., place orders by phone, pay through bank transfers,
and so on). While there is a loosely followed continuous supply agreement between SR
Farm and Uchumi, the farm would like to move to a stricter implemented supply contract.
Without the latter SR Farm is unlikely to make additional investments in its FF V
production for the domestic market (such as making the organizational investments

needed to optimize its harvest cycles to the needs of the supermarkets).
Type 3: Specialty Product Farms

The third type of supermarket-channel farmer, of minor importance at this point in time

(4% of suppliers), consists of specialty produce suppliers who have catered to more

127

sophisticated markets domestically or abroad. Here we find for example the producers of
pre-packaged, pre-cut vegetables for export.

An interesting example is provided by GD Farm. Started in 2000 as a supplier to
the airline and tourism industry, GD Farm is a 10 acre farm specialized in organic mixed
salad packs, although it produces various other organic FFV as well (e.g., strawberries).
Because all products are organic, produce items can be marketed for twice the price of
their conventional counterparts. The farm has 30 employees, is fully irrigated, has five
acres under shadow netting, has a washing and packing shed and complies with HACCP
standards as this is demanded by the airlines. In order to get its highly perishable salad
bags in good condition to customers, they use the cold chain system of a leading meat
processor, with whom they forged a partnership. Its salad bags are bar-coded and labeled
under their own brand name. In 2003, the airline and tourism industry on the one hand
and the domestic supermarkets on the other hand, each took about half of its supplies. Its
past growth has been financed by loans, retained earnings and credit by suppliers (e. g., of
irrigation equipment). The farm is developing various new products to extend its range
and plans to increase the acreage under production as well as make additional
investments in irrigation. The main attraction for supermarkets of having GD Farm as a
FFV supplier is that (1) it helps them complete both their line of value-added vegetable
packs and their line of organic F FV items, which is important in attracting the higher
income consumer segment; (2) it has the capacity (experience from other sectors, packing
shed, clean water, and so on) to wash, package, correctly weigh and label the product

(important aspects for these higher value-added produce items intended for the more

128

discerning high-income consumer segment); and (3) it has the capacity to supply these

produce items year-round (irrigation system).

Type 4: Market Diversijying Farms
The fourth type of supermarket-channel farmers consists of a varied group of farmers
whose main product focus is FFV for the domestic market, but for whom supermarkets
are not the main focus. These farmers have merely diversified their market by supplying
part of their output to supermarkets without showing much change in their farm practices
or turnover because of it. Of the supermarket-channel farmers, 31% fall into this
category. Here we find most of the fluit suppliers, which include (i) farms who have
recently shifted flom coffee (low prices) to bananas or from export crops (market
exclusion, e.g. due to inability to meet quality requirements in export markets) to
vegetables for the domestic market, (ii) medium-sized to large fluit growers and (iii) big
plantation farms such as Delmonte (13,000 acres of pineapples). For the larger producers
in this category, supermarkets are expected to have little or no impact for the next five to
ten years, as they represent only a minor part of the turnover of these producers.
However, small and medium sized farms in this category are expected over time to either
fully engage in the supermarket channel (thus becoming farmers of the first type) or to
exit this channel because the participation costs or the opportunity costs are too high.

F Farm is an example of the smaller farmers in this category of supermarket-
channel suppliers who finds himself at such a cross-road. Established in 1997, F Farm is
a 15 acre farm near Naivasha (85km west of Nairobi) which employs 4 full-time workers

and 10 casuals when needed. The farmer used to be in poultry, but exited this market

129

when prices were too low. He then shifted to FFV, growing a very wide variety of FF V
(13 different crops at the time of the interview, flom celery over broccoli to papaya).
Some of these items are sold to the leading supermarket chains (since 2000) in addition to
traditional buyers. For example, 80% of his spinach sales go to the supermarket while
20% goes to the main wholesale markets in Nairobi. Whatever cannot be marketed for
either lack of quality or lack of demand goes to waste.

For the supermarket, F Farm is a qualified supplier because it has an irrigation
system, its own transportation, a phone, and so on, indicating it can meet the
requirements for year-round supplies and short order-cycles (in the case of more
perishable leafy vegetables). The farmer likes selling to the supermarkets because they
are easy to supply to (less time-consuming), because the volumes are larger and because
the payment systems allows him to organize the farm’s accounts (e. g., worker payments).
The problem is that the orders he gets are too erratic (supermarkets telling him “call
tomorrow or the day after”), which makes it difficult to organize the harvesting. The
farmer also feels that receiving agents at supermarkets are not knowledgeable about
produce and that supermarkets take too long to pay him (up to a month). If this changes
and he gets more stable orders, he would further invest in the farm (especially irrigation)
and would likely become a dedicated supermarket-channel farmer. If things remain as
they are, the farmer will likely exit the supermarket channel, already indicating that

should prices for poultry get better, he would move back into that direction.

130

4.4 Cases of Smallholder Farmer Groups Supplying Supermarkets

Based on the estimated F FV turnover and distribution over the different types of
suppliers (essay 1), we estimate that around 1,000 smallholder farmers were part of the
supplier base of the supermarkets in 2003. These farmers supply (1) indirectly through
brokers/wholesalers (roughly 900 farmers) and (2) as (irregular) direct suppliers to
individual stores, mostly for highly perishable vegetables like leafy greens to up—country
branches (roughly 100 farmers). With supermarkets shifting away flom brokers to direct
supplies by farmers and moving to procurement centralization (essay 1), the importance
of smallholder producers in the supermarket channel is expected to dwindle to about 20%
of F FV supplies (their absolute number will be determined by the FFV turnover in
supermarkets as this is growing rapidly flom a small base).

Given this bleak scenario for smallholder producer participation in the
supermarket channel on the one hand and the rising importanCe of supermarkets and the
potential benefits of becoming a regular (listed) supplier to them (see section 5) on the
other hand, we briefly present five cases that demonstrate different strategies that may
allow for a greater participation by smallholder farmers in the supermarket channel.
These cases represent all the formats we could identify (based on interviews with
industry experts, including the superrnarkets’ F FV procurement managers) of smallholder
involvement other than through traditional-channel brokers or through direct supplies.
The types of suppliers presented in these cases were not included in the distribution of the
FFV suppliers presented earlier because they represent (in this early stage of supermarket

development) too small a flaction of the supermarket’s FFV supplies (<l%). Across the

131

five cases, three key success factors emerge: (1) a focus on products with a clear market
potential; (2) the catalyzing involvement of private or public organizations as marketing
facilitators with a commercial basis (i.e., intended to be sustainable without subsidies);
and (3) group formation amongst farmers. The case-information is mostly based on in-

depth interviews with key stakeholders, but also on secondary information sources.

Case I : Family Concern — An NGO Organizing Smallholder Farmers

Family Concern, a Kenyan NGO whose mission it is to combine development and
business objectives in building marketing linkages for small scale growers, is the lead
organization in a project that aims to facilitate the supply of traditional Aflican
vegetables (TAV)24 flom smallholder producers to domestic supermarkets in Kenya. The
lntemational Plant and Genetic Resources Institute (IPGRI) and the Kenya Agricultural
Research Institute (KARI) where involved as technical partners on the production side,
FARM Africa (a UK based NGO) was involved to provide expertise and financial
support and Uchumi was involved on the marketing side.

Based on consumer market studies, Family Concern had identified that (1) there
existed a high and growing demand for TAV, and (2) that this demand was not matched
by supplies. Uchumi had come to the same conclusion as it could not keep these
vegetables on the shelves and was trying (but not fast enough succeeding) in getting its
listed FFV suppliers to produce more. In 2003, Uchumi estimated that potential demand
for TAV was three times the actual sales of 100MT per month. When in 2003, Family

Concern approached Uchumi with the proposal to organize smallholder farmers for the

 

2" These are leafy vegetables such as black nightshade, spiderplant, cowpeas and amaranth.

132

production and marketing of TAV, the supermarket was very interested to get involved.
Family Concern and its partners then went to work, enlisting 50 smallholder producers in
the rural and peri-urban areas around Nairobi (with on average .25 acres available for
TAV) and organizing them in a sort of outgrower scheme to organically grow TAV under
well-defined good agricultural practices (e.g., with regard to use of clean water). Farmers
receive seed and technical assistance, are inspected with regard to their farm practices
and are, since 2004, marketing their produce through Family Concern, who is responsible
for transporting the bunches of TAV to Uchumi based on the latter’s order schedule. In
order to differentiate its product in the market place, Family Concern supports its product
with in-store advertising and plans to introduce its own label (African Delicacies).

The project benefits farmers as well as the supermarket. Farmers get assistance in
growing a cash-flow fiiendly product (ready for harvest after only 4-5 weeks, pest and
diseases resistant) for a reliable market, while Uchumi resolves (at least partially) one of
its key F FV supply shortages, expands its lines of organic and branded FFV items and
improves its image as it advertises its assistance to smallholder farmers to consumers in

its stores.

Case 2: HPHC — A Government-Owned Company Organizing Smallholder Farmers

Established in 2003, the Horticultural Produce Handling Company (HPHC) is a
govemment-owned company set up to own and commercialize an elaborate modern cold
chain inflastructure (financed by the Japan International Cooperation Agency JICA)
which is intended to improve the marketing system for horticultural produce from

smallholder producers. While initially managed by the parastatal Horticultural Crop

133

Development Authority (HCDA), government involvement is expected to be greatly
reduced over the period 2004-2005 (both in terms of ownership and management). An
initial focus on the EU export market has been replaced by a dominantly domestic and
East Aflica regional focus because of difficulties in accessing export markets. HPHC’s
facilities include seven satellite depots with pre-cooling units, a central warehouse in
Nairobi (near Jomo Kenyatta International Airport) with cold storage facilities and
insulated trucks of various capacities to maintain the cold chain flom collection point to
supermarket.

Under the stringent quality and safety requirements normally associated with the
export market, farmers produce and supply as groups of 15-40 outgrowers to a collection
center which has a charcoal cooler, a toilet and clean water. Selected farms have sizes
varying flom 0.125 to 2 acres and are located within a 5km radius of a collection center
(which is financed by the farmers themselves). Pre-cooled and insulated HPHC trucks
(2MT capacity) then pick the produce flom these collection points and take it to the
nearest satellite depot. Larger trucks (8MT) take the produce flom the satellite depots to
the Horticultural Centre in Nairobi and flom there to various buyers. Farmers get
planting, harvesting and spraying schedules and technical assistance flom HPHC as well
as buy their seeds from them. For this organization of the farmers and the cold chain
distribution, HPHC charged (in 2003) a low commission of 17% of the sales price
leading to higher prices for farmers (the 17% is far below the 65% of the wholesale price
going to brokers and wholesalers in the traditional channel). One of these buyers has been
Uchumi who bought Asian vegetables flom HPHC in 2003 (albeit for small volumes and

on an irregular basis).

134

HPHC has not started off as an instant success. Lack of marketing experience and
trading network partners have largely kept it out of export markets and in 2003 the
company was operating at only 5% of its capacity. However, its recent shift in market
focus to the domestic and regional market may fit well the strategic objectives and needs
of a supermarket like Uchumi. HPHC can offer the volumes and the food quality and
safety assurances that are key pillars in the FFV procurement system currently being
developed by the leading supermarkets in Kenya. It is therefore not surprising that in
2003 HPHC and Uchumi were exploring potential collaboration. If HPHC succeeds in
getting its marketing strategies right (e.g., by getting supermarkets like Uchumi and
others in the region such as South Africa’s Shoprite on its customer list) then its potential
impact is large (HPHC’s management indicated to us that they aim to get a 5% share of

the domestic FFV market by 2006).

Case 3: Iga Muka — A Farmer Group Assisted by a Private Sector Marketing Facilitator

Started in 1989, Iga Muka is a self-help group of about 30 smallholder farmers growing a
wide variety of FFV on their farms located on the slopes of Mt Kenya (some 200km flom
Nairobi, i.e., further away than most FFV suppliers for the Nairobi market). The farmers
group succeeded in linking up with Uchumi via the Kenya Agricultural Commodity
Exchange (KACE). Operational since 2000, KACE is a private sector firm that facilitates
linkages between sellers and buyers of agricultural commodities, in part by setting up
market information points which have small trading floors where buyers and sellers (like
Iga Muka and Uchumi) can interact. KACE gets its revenues flom selling market price

information on commodities (e.g., through SMS messages on farmer’s cellular phones)

135

and taking commissions on brokered sales as well as from donor support (e.g., flom
USAID and DF ID). Although it plans to overcome it, KACE was still dependent on
donor support in 2003.

In 2002, Iga Muka contacted KACE on the possibility of finding a market for its
strawberries, which are the group’s biggest income earner. Through an initial contact at a
KACE auction in Nairobi, Iga Muka placed a first order with Uchumi which then turned
into a longer term supply arrangement. The strawberries are harvested and collected three
times a week. In a small packing shed, the strawberries are graded according to size,
stage or ripening and appearance and packaged in plastic punnets of l/2kg. The product
is not labeled, other than with a barcode to comply with supermarket requirements, as the
costs of labeling are considered too high at this point. Iga Muka has a full-time grader on
its payroll at the collection point and has a marketing agent (representative) in Nairobi.
This agent collects the orders flom Uchumi (and other buyers). As the group does not
have its own transportation, it uses public transportation (mini-buses) to transport about
60 to 100kg of strawberries three times a week to Nairobi (200km of partially very rough
roads). Small volumes of about 60-80kg of strawberries per week, out of a maximum of
300kg, are intended for Uchumi, the rest is either marketed through KACE to non-
superrnarket buyers (e.g., institutions, greengrocers, and so on) or goes to waste.

While the group has succeeded in remaining in the supermarket channel, it is not
really equipped to supply supermarkets and has not been increasing its business because
profits are too low and access to credit is restricted. For example the group would like to
invest in small-scale processing equipment which would help solve the demand volatility

problem as well as make the grading more efficient (first grade to supermarkets, lower

136

grades for jam). However, it has not been able to build the capital flom retained earnings
while commercial loans are considered too risky. The absence of growth as well as
increasingly delayed payments by Uchumi (which was working through some financial
problems in 2003-2004) have lowered morale and undermined the internal stability of the
group which would have exited the supermarket channel already if it were not for its
relatively transparent accounting system and for its good organization (i.e., a constitution
that spells out a code of conduct for its members in addition to penalties for going against

the group’s bylaws).

Case 4: Sunripe — An Exporter with Smallholder Outgrowers Supplying Supermarkets
Sunripe, a family business founded in 1969 and in 2004 one of Kenya’s largest FFV
exporters, growing at an average annual rate of 20% over the last 15 years (Shah 2004),
produces 40 different high-value lines of fluits and vegetables flom its own farms as well
as flom 1,000 contracted smallholder producers. A (small) part of its output is sold to
Uchumi.

Sunripe is a change leader in Kenya’s export industry. In 1999 for example, it was
one of the first exporters in Kenya to start developing flesh-cut and washed & ready-to-
eat vegetable prepacks for the export market. Because (1) there are lower grades of
produce which fail to meet export standards and (2) there is sometimes oversupply if the
quantity contracted from farmers exceeds fluctuating demand from buyers in the export
market, exporter like Sunripe have excess produce for which they need alternative
markets. Around the same time (since 1997 and onward) supermarkets like Uchumi were

rapidly developing their FFV sections, continuously adding new items (especially in the

137

hyperrnarkets). Given this strategic fit between supply and demand, Sunripe started
supplying Uchumi with a number of different organically grown pre-packed vegetables
(e.g., pre-cut French beans, mixed stir-fly vegetables, and so on), providing an alternative
market for Sunripe and making a major contribution to both its line of organic products
and its line of pre-packed vegetable products for Uchumi. Another area where the two
companies are considering collaboration is with respect to the setting-up of quality and
food safety control systems for Uchumi’s F FV suppliers. Sunripe, which became
EurepGAP certified in October 2003, would be an excellent source of expertise in this
area, especially with respect to the implementation of traceability systems (for food
safety) amongst smallholder producers.

Sunripe’s 1,000 smallholder outgrowers, thus demonstrate yet another way in
which smallholder producers of FFV can be connected through non-traditional channels
to domestic supermarkets in Kenya. However, recent research indicates that the
implementation of increasingly stringent food safety standards imposed by EU
supermarkets on their FFV suppliers will likely lead to a declining role for smallholder

farmers in Kenya’s FFV export sector (Dolan and Humphrey 2000).

Case 5: B Farm - A Lead Farmer Organizing Smallholder Farmers

Another format under which smallholder growers could remain or become linked to
supermarkets is that of a lead farmers who organizes and buys flom a group of other
farmers. While this type of supplier is not yet important in the supplier base of the
supermarkets, it appears to be increasing in importance as several of these groups have

emerged in 2003-2004. B Farm, a 16 acre family farm located 25km flom Nairobi,

138

provides a good example here. This farm was started in 1992 and began supplying kale to
Uchumi in 2002. As the supermarket orders grew in size and regularity, the farmer
decided to source kale flom other farmers rather than to expand his own farm. By 2004,
he sourced flom 40 other farmers (mostly smallholders) who are as a group organized so
that the supermarket can be supplied with bunches of kale on a daily basis. For the
participating farmers, the key differences between selling to brokers or selling to B farm
is that (i) they now face a much more secure demand for kale and (ii) they get a better
price (as B Farm pays a premium to secure supplies). These are crucial differences for
smallholder farmers as risk reduction and higher profitability are important steps toward a
grth dynamic. B farm is also a retailer of farm inputs which further facilitates access to
inputs for the participating smallholder farmers and may be a precursor on the path to

interlinked supply contracts.

5. The Impact of Supermarkets in the Kale Supply Channel

5.1 Kale Supply Channels in Kenya

Kale Production Characteristics

There are two key characteristics of kale production in Kenya that are important for the
present analysis. First, most of the kale production in Kenya takes place under rain-fed
conditions, leading to strong supply and price seasonality (annual production cycles are
tuned to a short rains and a long rains season). This implies that for supermarkets, who
want year-round supplies, kale must be ordered flom farmers who have irrigation and a

year-round source of water. Second, there are no specific, formal quality or food safety

139

standards in the domestic market (in the traditional channel nor in the supermarket
channel) but quality, mostly determined by fleshness, leaf size and color and the degree
of damage due to pests or other decay, does play a role in price determination. For the
supermarkets who want to differentiate themselves on quality, only the best quality of
F FV will be accepted. This implies that farmers supplying supermarkets must be able to
control quality (e.g., through pest management, by being located close to Nairobi so the

highly perishable kale arrives flesh, and so on).

Kale Supply Chains

The entry of supermarkets in the kale supply chain leads to supply channel integration.
Kale farmers in Kenya have potentially four market options, depending on their
capacities: (1) they sell to brokers flom their farm; (2) they sell to brokers near the main
road; (3) they sell to wholesalers in main urban areas; and (4) they sell directly to retailers
or institutions. Which option is chosen is mostly determined by the transportation
capacity of the farmer and his or her willingness to take risks (the more downstream the
farmer takes the produce, the higher the price, but also the higher the transportation costs
and the higher the price risks). Given that most traditional-channel farmers are risk-
averse smallholder producers without (motorized) transportation, most farmers in this
channel sell to brokers at the farm or near the main road (rural collection market). For
example, in our farmer sample we found that of the Lari farmers (at 60km flom Nairobi),
52% sell kale flom the farm, 38% take it to the main road and 10% take it to wholesale
markets in Nairobi. Supermarket-channel farms on the other hand are larger and have

their own transportation and are capable of taking their produce direct to the

140

supermarkets. Figure 1 compares the traditional market channel with the supermarket
channel for kale, indicating the distribution of the marketing margin over the supply
chain’s successive economic agents, as well as an indication of the size (in terms of sales
volume) of these economic agents for the traditional .channel. Figure 1 indicates that the
shift flom the standard traditional supply chain to the supermarket supply chain for kale
implies a vertical integration, with the farmer integrating the broker and part of the
wholesale function and capturing a far greater part of the marketing margin. This extends
earlier research on the applicability of market channel theory in Kenya to the supermarket
channel. Dijkstra (2001) showed how channel theory (i.e., vertical channel disintegration
(i.e., the number of successive distinct business entities in the supply chain) is negatively
correlated with the size of producers and retailers) could explain long supply channels for
FFV in Kenya. Here we find the mirror image: supermarkets (as large retailers) and

supermarket-channel farmers (as larger producers) lead to short (direct) supply channels.

141

Figure 6: Marketing Margins in Kale Supply Chains

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Traditional Market Channel Supermarket Channel

Tfiesdgiime £353; / 11% , 17% / $3532..

40 Farmer / 17% , Fariner
4,500 Broker _——-—-—-‘ 29% 57%

 

 

 

 

 

 

 

 

 

 

 

 

 

400 Wholesaler
\ 23%
50 Retailer ,_ / Retailer
\ 20% 26%

Consumer Consumer

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: margin to input suppliers in traditional chain assumes farmers buy seeds and manure, rather than use
their own.

Source: author’s farmer survey, interviews with 15 channel participants, including brokers, wholesalers and
retailers and price-measurements throughout the chain.

Traditional Kale Supply Chains

Traditional kale supply chains provide marketing solutions to smallholder farmers, but
are characterized by a small percentage of the marketing margin captured by the farmer,
high transaction costs (produce changes ownerships many times) and brokers with high
market power (they trade the largest volumes and are most knowledgeable about prices at
the farm and in the wholesale markets). In the standard traditional supply chain for kale, a

farmer stuffs the harvested kale leafs in bags of 75kg and takes them by donkey cart to an

142

open air collection market near the main road to Nairobi”. At this rural market,
independently operating brokers arrive with a pick-up or a larger lorry. They negotiate
the prices with farmers and, once a price is agreed upon, they pay loaders to load the
lorry to maximum capacity. The broker then takes the kale bags to one of the wholesale
markets in Nairobi, pays a fee to enter the market, and starts negotiating prices again, this
time with several wholesalers“. Once a price is agreed upon off-loaders are paid by the
buyer to unload the bags of kale. It takes about 10 wholesalers to buy a whole lorry-load
of kale and so it can take brokers the better part of a day to sell it all. Wholesalers then
take leafs flom the bags and bunch them in bunches of approximately 650gr. In the next
step, small, traditional retailers such as market stalls, kiosks or street hawkers come to the
wholesale market where prices (this time per bunch) are once more negotiated. Retailers
then pay cart pullers to bring the bought produce to their retail outlet. These retailers then
break up the bunches into smaller bunches of maybe 200gr and offer them for sale to the
end-consumer, who again will negotiate the price.

While quite challenging for the resource-poor, risk-averse smallholder producers
(who form the bulk of the producers in the traditional channel), there are variations to the
above supply chain structure which allow farmers to capture more of the marketing
margin and reach an intermediate step to selling to supermarkets. Farmers may form
small groups whose combined harvest is sufficiently large to make it economical to rent a
truck and take the produce directly to the wholesale market or even to retailers (some of

the ad hoc supplies of kale to supermarkets come flom smallholder farmers this way).

 

2’ Alternatively, the farmer may sell directly off-farm to a broker who was directed to the farm by agents
hired by the broker. Agents are local people who are very knowledgeable about which farmers have a ready
harvest. Or the farmer may pay another farmer to take her or his bags to the road-side market.

26 Alternatively, a small group of wholesalers may join forces and hire a transporter to go to the rural
market to buy a number of bags of kale or they place a joint order with an established broker.

143

Farmers may also bunch the kale leafs themselves in order to capture the value-added.
When bunches are made closer to the farm, they are usually transported in extra large

bags (approximately 200kg).

The Supermarket Kale Supply Chain

At the other extreme is the supermarket supply chain for kale, which is relative to the
traditional supply channel shorter, more efficient flom a transaction cost point of view,
allows the farmer to capture a large portion of the marketing margin (farmer and
supermarket each capture a part of the wholesaler’s margin) and controlled by the
supermarket in stead of by the broker (who disappears). The supermarket calls the farmer
in the afternoon with the order for the next day, indicating also if the price has changed
since the last order and if so by how much. Early in the morning of the next day, the
farmer puts his entire workforce (if needed supplemented with casual labor) on the kale
harvest, and in about 2 hours the farm workers have picked and bunched enough leafs to
fill the order and have loaded them directly on a truck (no bags), covered with canvas.
The farm’s driver then takes the kale to the supermarket’s reception bay, has the delivery-
book filled out and returns to the farm less then an hour after delivery. Supermarket
employees place the kale bunches in the FFV section. The farmer is paid bi-weekly (in
principle) for the cumulative supply over that time period. Given the wide variety of
produce items grown by supermarket-channel farmers, they usually supply more than one

item per delivery to the supermarkets.

144

5.2 Determinants of Farmer Participation in the Supermarket Channel

In this section we want to assess how farm characteristics determine the participation of
the farm in the supermarket channel for kale (Y). Since negative dependent variables are
not possible and assuming non-linear effects of the explanatory variables, we modeled

this channel adoption decision as a probit model. The model takes on the following form:

Prob (Y=l IX) = G(Bo + X13),

where G is the standard normal cumulative distribution function. In their most general
form, adoption functions contain the following five categories of explanatory variables:
prices of inputs and outputs, risk factors, quasi-fixed capital and shift factors. Prices of
inputs and outputs were not directly included as they are implicit in the channel choice
and further determined by the size and location of the farm. The risk factor and quasi-
fixed capital explanatory variables we want to include here capture risk-sensitivity (land
ownership), land capital (size of the farm), access to financial capital (size of land and
ownership, education), human capital (age, education, gender) and physical capital
(presence of an irrigation system). Each of these explanatory variables is hypothesized to
ceteris paribus increase the probability of adoption of the supermarket channel. With
regard to gender it is hypothesized that men are more likely to enter the supermarket
channel because they (1) are assumed to have better access to the required production
factors and (2) tend to get more involved when the transactions become more formal,

sizeable and rewarding (Dolan 2001). Two variables normally presented in adoption

145

models, household size (quasi-fixed capital) and location (shift factor), were omitted
here. Most of the supermarket-channel farmers rely for the greater part on (often large
numbers of) hired employees, making size of the household (indicating the availability of
family labor) not a very meaningful variable. Location was left out the model because the
traditional-channel farmers were selected flom only two nearby divisions so that for this
sub-sample there was too little variation over the location variable. One further departure
flom the standard adoption model is that for the size of the farm, land ownership and the
presence (or absence) of an irrigation system we used the 1999 situation. The 1999 data
reflect the position of the farms at the time when they became suppliers to supermarkets
(all the supermarket-channel farmers became suppliers to supermarkets around or shortly
after 1999). If we would have used the current situation (e.g., the size of the farm in
2004) then there could potentially be an endogeneity problem as farm-size could well
have been influenced by supermarket channel participation. Based on the above, the

implementation model has the following determinant variables (X):

(a) the percentage of the farm size that is owned in 1999 (percent owned 99);

(b) the size of the farm in acres in 1999 (size 99);

(c) the number of years of schooling of the head of the farm (education of head);
((1) the age in years of the head of the farm (age of head);

(e) the gender of the head of the farm (gender of head); and

(f) a dummy variable indicating if the farm had an irrigation system (sprinkler or

drip) in 1999 (irrigation 99).

146

Based on maximum likelihood estimation (MLE), Table 17 presents the probit estimators

(B i) of the above model. Table 17 indicates that the results are as hypothesized, namely,

the probability of a farm participating in the supermarket channel increases as the farm (i)
is larger and (ii) has a drip or overhead irrigation. The marginal effect indicates that, for
the average farm (in terms of the independent variables), having one more acre of land
(i.e., a relatively large 10% increase relative to the average size of 10 acres) increased the
probability that the farm will participate in the supermarket channel with nearly 12%
while having a drip or overhead irrigation system increases this probability with a
dramatic 46%. The latter indicates that having irrigation is a critical capital requirement
for farms who want to become FFV suppliers to supermarkets. Education, age, gender
and land-ownership did not have ceteris paribus a statistically significant impact on the

probability of a farm participating in the supermarket channel in our sample.

Table 17: Determinants of Farmer Adoption of the Supermarket Channel (Probit
Results)

 

P(Supplies kale to Marginal effect

Independent variables supermarkets) (s.e.)

 

 

Size 99 0.30 (0.157)* +11.8%
Education head 0.19 (0.164) -
Percent owned 99 -0.01 (0.014) -
Irrigation 99 1.35 (0746)“ +46%
Gender of head 0.55 (1.078) -
Age of head 0.02 (0.031) -
Constant -5.76 (3.930) -
No. of observations 57

(Pseudo) R-square 0.7621

 

Notes: "' = significant at the 10% level. Marginal effect measured at the mean
levels of the determinant variables.
Source: author’s farmer survey 2004

147

5.3 Supermarket Channel Participation and the Farmer’s Production Technology

In this section we want to compare the production technologies used by supermarket-
channel farmers and traditional-channel farmers. In the micro-economic theory of the
firm, technology is represented by a production function which reflects the
technological relation that exists between any particular combination of inputs and the
resulting levels of outputs q (Sadoulet and de Janvry 1995). These inputs consist of an
X vector of variable inputs (e.g., fertilizer) and a Z vector of quasi-fixed inputs (e.g.,
land). Various types of production functions exist. Here we select the Cobb-Douglas
production function which is the most used functional form for the analysis of farm
efficiency (Battese, 1992; Bravo-Ureta and Pinheiro, 1993). This production function
takes on the following form:

q=Ax°‘zB

Where a and [3 represent the elasticity of production for the inputs with respect to the
output, i.e., the percentage change in output for a 1% change in the input keeping all
other inputs fixed (aqax/Oqu, Oth/OZtQ). By selecting land, labor and fertilizer as the
inputs and taking the natural logs, we get the following two production functions which

we want to estimate and compare:

Supermarket-channel farmers: lnoutput = A1+ afihrlabor + Bfilnland + yfi'lnfertil + e; (l)

Traditional-channel farmers: lnoutput = A2 + (12*lnlabor + [32*lnland + yflnfertil + e; (2)

148

Whereby lnoutput is defined as the natural log of output measured as kg of kale produced
by the farmer flom the current acreage over 1 cycle; lnlabor is the natural log of the
number of full work-days used for land preparation, planting and weeding of kale; Inland
is the natural log of the number of acre-months used for kale growing (we want to take
into account a harvest cycle length which over the sample varies flom 2 to 12 months”);
and lnfertil is the natural log of the number of kg of fertilizer applied to the current
acreage over 1 cycle (summarized over the various types of fertilizer which in terms of
value per kg do not differ widely). The last term, e, is the estimated error (disturbance)

which contains the unobserved explanatory factors.

Chow Test

A method often used in econometrics to test for differences in regression functions across
groups is the Chow test (Wooldridge 2000). The Chow statistic tests the null hypothesis
that the intercept and all the coefficients are the same across the groups. For equations (1)

and (2) above we get:

H0: A1=A2, al= (12,131: 132.71: 72

The Chow test involves estimating a restricted model (the four restrictions under Ho) and

an unrestricted model (under the alternative hypothesis that Ho does not hold). The

following F statistic must be calculated:

 

27 Kale leafs can be harvested flom the same plants at regular intervals over periods stretching up to 12
months, although the normal harvest cycle is about three months (East Aflica Seed Co. 2002).

149

Fk+1, n-2k-2 = [(SSR, — SSR1 - SSR2)/(SSR1+SSR2)] . [(n-2k-2)/(k+l)],

where (i) SSR are the sum of squared residuals obtained (through ordinary least squares
estimation) for the restricted model (using all observations) and for two separate
regressions, one for each group (using only the observations for one group at a time), (ii)
k is the number of coefficients to be estimated and (iii) 11 is the overall number of
observations. The value of the F -statistic here is 1.891 with a corresponding p-value of
0.129. This means we can reject H0 at the 15% level. The test indicates that there is a
statistically significant difference (at the 15% level) between the production functions of

supermarket-channel farmers and traditional-channel farmers.

Estimation of the Production Functions Corrected for Self-Selection Bias

The distribution of the farmers over the two groups (supermarket channel, traditional
channel) is not random but rather the outcome of a self-selection process. In order to
control for this (potential) selectivity bias, we use Heckit’s two-stage method. In the first
step, this method calculates the inverse Mill’s ratio (3.) for each observation (via a probit
model for channel choice) which then in the second step is added in to the substantial

model (the production function) as a control variable. The models now become:

Supermarket-channel farmers:
lnoutput= A] + afilnlabor + Bfilnland + yfl'lnfertil + 61*)» + e1 (3)
Traditional-channel farmers:

lnoutput = A2 + aflnlabor + 132*lnland + yflnfertil + 52*}. + e2 (4)

150

Table 18 lists the estimated parameters for both models. The 7» regressor is significant for
equation 4 which means its presence corrects for a self-selection bias. We further find
that both models are significant and that the signs of the coefficients are as expected,
namely output increases as, ceteris paribus, more land or more fertilizer are used. Labor

was found not to have a statistically significant effect in either model.

Table 18: Heckit Two-Stage Estimates of the Cobb-Douglas Production Functions

 

 

Dependent variable: lnoutput Supermarket farmers Traditional market farmers
Independent variables Coefficient (s.e.) Coefficient (s.e.)
Constant 6.8237 (1.1484)*“ 6.6034 (0.8699)"*
Lnland 0.8681 (0.3133)**"‘ 0.5794 (0.2453)”
Lnlabor -0.3254 (0.2527) 0.0446 (0.2388)
Lnfertil 0.2407 (0.1070)" 0.2030 (0.1553)
Mill’s lambda 0.1952 (0.2471) 0.9889 (05379)"
Wald Chi2 107.3 23.25
Significance 0.0000 0.0000

 

Notes: ** = significant at the 5% level, *** significant at the 1% level.
Source: author’s farmer survey 2004.

Table 19 compares the marginal product values for the significant coefficients of table 6.
The marginal product value is calculated by multiplying the marginal productivity of a
factor x, (6q/6x,=a . q /xi) with the market price of the output. Several observations can be
made. First, the data indicate that the marginal product value (MPV) differs greatly
between the two groups of farmers: in absolute terms the MPV is larger for the well-
capitalized supermarket-channel farmers while as a relative measure (relative to gross
revenue) it is larger for the traditional-channel farmers who capture only a small
percentage of the marketing margin. Where one additional acre-month of land would
increase revenue flom the sale of kale with Ksh3,217 for traditional-channel farmers (i.e.,

roughly 10% of their average total revenue per acre), it would increase revenues for

151

supermarket-channel farmers with Ksh8,073 (i.e., roughly 5% of their average total
revenue per acre). Second, each of calculated marginal product values is larger than the
corresponding factor cost, indicating that these factors are used below the optimal
quantity for both groups of farmers. For naditionalechannel farmers, this may indicate
that land access is constrained by a limited access to land (high population density). For
supermarket-channel farmers, who on average only use 70% of their land, the bottleneck
is more likely to be found in the limited access to capital for irrigation system expansion.
Third, the average land productivity and the average labor productivity are respectively
59% and 73% higher for supermarket-channel farmers than for traditional-channel
farmers. This results flom the more capital intensive production methods used by
supermarket-channel farmers (more variable inputs like fertilizer or chemicals and more
quasi-fixed inputs per unit of labor and per unit of land). Relative to daily wages (which
are Ksh150/day and Ksh120/day for supermarket-channel and traditional-channel farmers
respectively) average labor productivity is four times the wage rate for supermarket-
channel farmers and three times the wage rate for traditional-market farmers. This
extends to the supermarket-led part of the FFV supply system, the findings of Carter and
Wiebe (1990) that higher access to capital (in this case of the supermarket-channel

farmers) impacts the agrarian structure and productivity in Kenya.

152

Table 19: Input Use and Farm Efficiency

 

 

 

 

Supermarket farmers Traditional farmers

Input MPV FC MPV FC
Land (acre-months) 8,073Ksh > 667Ksh 3,217Ksh > 1,083Ksh
Fertilizer (kg) 56Ksh > 28Ksh

Productivity Measures
Labor productrvrty (average 60 5Ksh 3 49Ksh
output value per work day)
Land productrvrty (average 1 0,2 6 4Ksh 6,43 7Ksh

output value per acre-month)

Notes: marginal product value (MPV) valued at farm-gate price of Ksh3.6/kg. Factor costs (PC)
are sub-sample averages.
Source: author’s farmer survey 2004.

5.4 Net Income Effect of Supplying Supermarkets

In this section we assess whether or not there is a positive net income effect for farmers if
they supply kale to supermarkets rather than to traditional buyers. Table 20 provides a
gross margin analysis for the two types of farmers. The top half in this table, which looks
at production, allows for a straight forward comparison between supermarket-channel
and traditional-channel farmers. However, the marketing part forced us to make a choice.
The supply channel for supermarkets has a particular given structure, with all farmers
taking their harvest directly to the supermarket. Traditional-channel farmers can however
supply to brokers, wholesalers or retailers (other than supermarkets). Furthermore, either
type of farmer may sell to a wider set of buyers at different links in the supply chain. For
example, a supermarket-channel farmer may sell his highest quality grade to
supermarkets and the lower grades to wholesalers. Or a traditional-channel farmer may
sell to a broker at the farm for one harvest, but collaborate with other farmers to take the

next week’s harvest to a wholesaler. However, in most cases, both types of farmers sell

153

most, if not all, of their harvest to a single type of buyer. In table 20 we opted to compare
the most extreme (but also most common) marketing choices, namely supermarket-
channel farmers sell (nearly) 100% of their kale to supermarkets (applies to 60% of the
farmers) and traditional-channel farmers sell (nearly) 100% of their kale to a broker at the
farm (applies to 46% of the farmers).

Table 20 indicates that there are substantial differences between the two groups of
farmers. On average, supermarket-channel farmers use about twice the amount of inputs
(fertilizer, manure, chemicals) traditional retailers use, but pay less per unit as they buy
larger volumes. Supermarket-channel farmers use less seed, but seed of a higher quality
(traditional-channel farmers produce their own seed which is cheap and of lower quality).
The variable production costs incurred by supermarket-channel farmers further include
the costs for tractor rental and energy to operate an irrigation system. Traditional-channel
farmers use more labor per acre, mostly because there is an abundance of family labor
relative to the small farm sizes. Wages for hired labor are higher in the supermarket
channel than in the traditional channel. (Family labor was valued at the same rate as hired
labor in table 20.) Given the higher input levels, yields per acre are higher for
supermarket-channel farmers”.

When we translate these farm practices into the production cost per kg of kale, we

see that there is almost no difference between the two groups of farmers, with both types

of farmers producing kale at about 3.5Ksh/kg on average. From this comparable starting

 

2" Yields per acre vary greatly because of differences in variety, harvest cycle length, soil quality, climate
and farm practices, but are estimated at 6MT per acre on average (MoARD 2002) while yields up to 15MT
per acre are possible (for comparison, kale yields in the US can go up to 18MT per acre; Oregon State
University 2002). The relatively high yield found here for traditional-channel farmers may be due to the
specific selection of our sample (a key kale production area for the Nairobi market) or to the farmer’s
overestimation of the quantity harvested and/or underestimation of the area under kale.

154

point at the production level, the two paths diverge. Traditional-channel farmers incur
limited marketing costs (only harvesting and bagging), but also receive a low price flom
brokers at the farm-gate which allows them to break even at best. Supermarket-channel
farmers on the other hand also incur transportation costs, but receive a price which is
more than three times the farm-gate price, resulting in a gross profit of about 40%.

Let us now return to the net income effect of supplying supermarkets. Before
selling to supermarkets, most current supermarket-channel farmers already had the
capacity to market their production to buyers in Nairobi. Selling to supermarkets, who
buy larger volumes per delivery, lowers the farmer’s transaction costs (several farmers
told us that the time spend in selling was reduced flom a day to an hour) and pays them a
higher price (about 10-20% higher), and thus has a strong positive net income effect for
these farmers. For individual traditional-channel farmers, shifting flom a broker at the
gate to selling to supermarkets would not have a positive net income effect as the
transportation costs would be prohibitively high (given the small volumes). Table 21
indicates how the supply capacity differs between the two groups of farmers. When
harvest cycle length and size of the farm area under kale are taken into account,
supermarket-channel farmers can supply more than five times the volume of traditional-
channel farmers. This large difference in supply capacity (and the fact that supermarket-
channel farmers grow a wider variety of crops) allows for far more efficient

transportation.

155

Table 20: Gross Margin Analysis Kale Supermarket vs. Traditional-channel
farmers

 

Data are per acre

100% to supermarkets

Farmers supplying

Kiambu farmers supplying
100% to brokers at the farm

 

 

 

 

 

Unit Cost % Unit Cost %
Line Items1 Unit Units Cost Total of . Units Cost Total of

(Kg (Ksh) Rev. (Ksh) (Ksh) Rev.
Seed " Kg 0.4 1280 512 0.3 1.1 790 869 2.6
Inorganic fertilizer Kg 314 28.0 8,792 5.7 177 28.8 5,098 15.1
Manure ""' MT 6.6 1,300 8,580 5.6 3.3 2,100 6,930 20.6
Chemicals " L 3.3 1,090 3,597 2.3 1.6 1,135 1,816 5.4
Cost Irrigation (var.) Month 5.4 2,200 11,880 7.7 0 na 0 0
Tractor rental Acre 1 2,000 2,000 1.3 0 na 0 0
Total inputs 35,361 23.0 14,713 43. 7
Labor by activity
Land preparation Mds 17 150 2,550 1.7 28 120 3,360 10.0
Planting Mds 12 150 1,800 1.2 19 120 2,280 6.8
Weeding Mds 55 150 8,250 5.4 85 120 10,200 30.3
Labor by Source
Family labor "* Mds 8 150 1200 0.8 120 120 14400 42.8
Hired labor "* Mds 76 150 11400 7.4 12 120 1440 4.3
Total Labor Cost Mds 84 150 12,600 8.2 132 120 15,840 47.1
Total Product. Cost Kg 12,800 3.7 47,961 31.2 9,350 3.3 30,533 0.8
Harvesting Mds 77 170 13,090 8.5 26 120 3,120 9.3
Take & Sell Market Mds 29 300 8,700 5.7 0 Na 0 0
Rope (bag or bunch) Pce 19,700 0.08 1,576 1.0 125 2 250 0.7
Bags Bag Na Na 0 0 125 20 2,500 7.4
Transport (var.) Km 1,700 10 17,000 1 1.1 0 Na 0 0
Phone Cost Na Na Na 390 0.3 Na ' Na 0 0
Total Market. Cost Kg 12,800 3.2 40,756 26.5 9,350 0.6 5,870 17.4
Total Cost 6.9 88,717 57.8 3.9 36,423 108
Total Revenue 12.0 153,600 100.0 3.6 33,660 100
Gross Profit 5.1 64,883 42.2 -0.3 -2,763 -8.2
Gross Profit (excl.
family labor) 5.2 66,083 43.0 1.2 1 1,637 34.6
Yield 12.8 MT/acre 9.4 MT/acre

 

Notes: (1) significance for production line items: "' = significant at the 10% level , " = significant at the
5% level, "" significant at the 1% level; (2) numbers are indicative only as they are based on the

farmer’s estimations, not direct measurement; (3) prices used are weighed sample averages for both

groups of farmers; (4) weight conversions used: 1 bag=75kg, 1 bunch=0.65kg (author’s field

measurements).

Source: author’s farmer survey 2004.

156

Table 21: Supply Capacity Supermarket vs. Traditional-channel farmers

 

 

Traditional Supermarket Difference
Farmers Farmers
Yield 1 (MT/acre) 9.4 12.8 +37%
Harvest Cycle Length (months) 5.1 4.4
Yield 2 (kg/acre, month) 1,843 ~ 2,909 +59%
Average Land Under Kale (acres) 0.9 3.6
Supply Capacity (bunches/week) 415 2,600 +527%

 

Source: author’s farmer survey 2004.

However, if traditional-channel farmers would join together to market their output,
supplying to supermarkets would have a dramatic effect on their net income, as it would
allow them to capture the 40% gross profit margin now enjoyed by the supermarket-
channel farmers. Various factors hinder the emergence of marketing groups amongst
traditional-channel farmers. First, there is a dire lack of experience with marketing
produce (in fact, most farmers dislike to be engaged in marketing activities). Second,
risk-averseness deters traditional-channel farmers flom selling to more downstream
buyers. On the one hand, selling to wholesalers exposes farmers to high price fluctuation
risks which may average out for a broker supplying daily, but which could financially
cripple a group of smallholders who could maybe market once a month. On the other
hand, selling to supermarkets, which could provide more stable prices, implies that
farmers have access to affordable credit and are willing to risk taking out a loan to invest
in the irrigation system required by supermarkets. Furthermore, selling to supermarkets,
who at times take up to a month to make payments, will often also exceed the working

capital capacities of cash-strapped farmers.

157

5.5 Supermarkets and Farmer Growth Dynamics

Confirming the financial benefit of increased net income for farmers who can access the
supermarket channel as indicated in the previous section, 93% of these farmers said that
supplying to supermarkets has affected their profitability favorably. However,
supermarkets also provide other benefits which are even more important to farmers than a
higher price. While 34% of the supermarket-channel farmers say that the higher price is a
key reason for selling to supermarkets, 46% say that the ease of selling to supermarkets is
the key attraction. Consider a farmer in the traditional channel. Uninformed about the
price of the day (maybe reacting to the previous day’s price), the farmer decides to
harvest (maybe because there was no longer time to postpone it or because of an urgent
need for cash) and takes her or his produce to the rural market. Many farmers and few
brokers may show up at this market, leading to plummeting prices, which the farmer has
to accept, even if they are below production cost, as the opportunity cost of the kale is
close to zero (animal feed). Even within a given day, prices can fluctuate strongly,
making the marketing process much like a gamble. Now consider a supermarket-channel
farmer. Since supplies are made to order, the farmer is certain, before harvesting, that the
sale will take place and at what price. Prices paid by supermarkets change step-wise,
remaining constant over longer time intervals, a stability which is desired by both the
supermarket and the farmer. These orders are also coming in throughout the year with
some reliability (at least for the growing group of F FV suppliers who get long term

supply agreements). While supermarkets and their FFV suppliers start to build up long

158

term relationships, 95% of the traditional-channel farmers are selling in spot markets with
buyers varying all the time.

The combination of higher net incomes and greater stability in volumes and prices
in the supermarket channel, have created a strong growth dynamic. Of the farmers
supplying kale to supermarkets, 75% said they increased production in response to
supermarket demand. Supermarket-channel farmers have kept investing in their farm,
resulting in a strong growth: the average farmed acreage of supermarket-channel farmers
increased by 104% over 1999-2004, compared to by only 10% for traditional-channel
farmers. In search for growth, the single most important input access constraint faced by
supermarket and traditional-channel farmers alike is access to credit (Table 22). It is in
this context meaningful that 44% of the supermarket-channel farmers believe that their
status as a supplier to supermarkets (which gives them formal proof of a steady income
flow) has increased their access to credit. However, increased access to credit does not
necessarily mean affordable credit. Commercial credit is expensive (IS-20% interest
rate), while government supported loans (10% interest rate) are, according to the farmers,
too difficult to obtain because of slow and selective bureaucratic procedures. This
probably explains why even for supermarket-channel farmers access to credit remains a
key constraint. Nevertheless, growing as they are, the current list of supermarket-channel
farmers are likely to be able to follow growing demand for FF V by supermarkets for the
next five years: 71% of these farmers state that, should supermarkets ask them to, they
could double their current supply of kale within one (five month) cycle (92% of them by

increasing production, 8% by sourcing from other farmers).

159

Table 22: Farmers’ Access Constraints to Key Inputs

 

 

% of farmers indicatingthat: Traditional Farmers Supermarket Farmers
access to credit is a problem 88% 85%
access to inputs is a problem 43% 16%
access to land is a problem 8% 19%
access to labor is a problem 3% . 8%

 

Source: author’s farmer survey 2003.

6. Summary, Conclusions and Recommendations

Our research has shown that the rise of supermarkets has given rise to a new group of
farmers in the domestic FFV market. The key suppliers of FFV to supermarkets are
mostly recently established, medium-sized farms (10-40 acres) managed by well-
educated farmers and specialized in supplying supermarkets. Around 40% of
supermarket-channel farmers fell in this category in 2004 and it is expected that these
farmers will become dominant in the supplier base over the next five years. In terms of
their capital, supermarket-channel farmers differ starkly flom traditional-channel farmers.
Especially size and the presence of an irrigation system were found to be critical
determinants of participation in the supermarket channel. This is so because it addresses
the current F FV procurement priorities for supermarkets in Kenya: finding farmers who
can supply large volumes year-round. The fact that nearly all supermarket-channel
farmers have the capacity to supply large volumes, have one or more transportation
vehicles, an irrigation system, a packing shed, a cellular phone, and so on, points to the
presence of a threshold capital vector which farmers must have in order to enter the
supermarket channel. Only 25% of the farmers supplying supermarkets directly can be

classified as smallholders (less than 10 acres) and while we gave several examples of

160

groups of smallholders who, assisted by market facilitators, are linked into the
supermarket channel, the importance of such smallholder groups as direct suppliers of
F FV to supermarkets is minimal. We further found that amongst kale suppliers to
supermarkets, women are less frequently the head of the. farm (1/6 in stead of 1/3 of the
farms as is the case for traditional-channel farmers) and make up a smaller percentage of
the employed workforce (a third in stead of half).

Farms supplying kale to supermarkets have adopted capital intensive
technologies. They use more quasi-fixed capital (e.g., irrigation systems) and apply (per
acre) twice the amount of fertilizer, manure and chemicals used by traditional-channel
farmers outside the supermarket channel. Consequently, average land and labor
productivity are about 60-70% higher for supermarket farms. Although supermarket-
channel farmers include few smallholder producers (relative to the traditional channel)
and have a lower labor-to-land ratio, they employ many hired workers. On average there
are 13 full-time workers per supermarket farm and 80% of these are hired laborers. The
overall effect on labor is about a 20% reduction per acre. Furthermore, commensurate
with their higher productivity, these hired farmed workers are paid a 15% higher wage on
average. Our analysis further shows that, for kale, overall productivity of supermarket-
channel farmers and traditional-channel farmers is similar as both have nearly the same
production cost per kg of kale. Therefore, on the production side, the essential limitation
of traditional producers (85% of which are smallholders) is not their production
efficiency but their lack of scale and inability to produce year round (in the absence of

modern irrigation systems).

161

An even more important difference between supermarket-channel farmers and
traditional-channel farmers is their marketing capacity. Most (individual) traditional-
channel farmers do not have their own (motorized) transportation and lack the scale to
rent it. Therefore these farmers are forced to sell to brokers who come to the farms at
prices that allow them to break even at best. On the other hand, many supermarket-
channel farmers were already supplying their produce to wholesalers or retailers in
Nairobi and could easily make the switch to supplying supermarkets. The 10-20% higher
prices paid by supermarkets (relative to wholesalers) give supermarket-channel farmers a
healthy gross profit margin of 40% (for kale).

If smallholder farmers could achieve the same transportation efficiency as
supermarket-channel farmers (by marketing in groups) and could supply direct to
supermarkets, they would get three times the price they currently get flom brokers at the
farm, capturing the same 40% gross profits currently enjoyed by the supermarket-channel
farmers. Furthermore, supermarket-channel farmers indicated that the benefits of market
risk reduction (more stable prices and volumes) and reduced transaction costs are even
more important than the higher price. The combination of higher gross profits and a
stable long term trading relationship has been a powerful determinant of a strong growth
dynamic amongst supermarket-channel farmers. The latter have, on average over the last
five years, doubled the size of their operations (where the traditional-channel farmers in
this study only increased theirs with 10%). This strong growth further indicates that the
current supermarket-channel farmers will likely keep pace with the growth of

supermarkets and even increase their share of the supermarkets’ FFV supplies.

162

Development Policy and Program Recommendations

For development policies and programs, the key insight revealed by studies on the rise of
supermarkets in developing countries (like this one), is that supermarkets are changing
the structure of the food system whether directly through their actions or through the
competitive response they provoke in the traditional system, and that they are doing so
fast. Those (farmers) who adapt (fast) will benefit flom this change. Therefore, if we
want to help smallholders in this new context, whether by helping them to link up with
supermarkets or by making traditional channels more competitive, a similar set of
recommendations can be put forward. The following three recommendations will help the
design of development programs in taking the new reality of retailer driven food systems
in developing countries into account.

First, assistance programs must focus on complete supply chains, not individual
agents, not parts of a chain. It is not enough to merely start flom the market. There is a
need to design the whole chain from the beginning. As we have shown in this paper,
farmers need to meet a vector of criteria in order to gain access to supermarkets which
cannot be addressed by assistance programs targeting an isolated problem in a particular
industry (horizontal level), such as developing a higher yielding variety for farmers
independent of other considerations regarding the structure of the supply chain.
Essentially, this implies multiple partners to be involved in the development program,
especially retailers. The design of the supply chains may further imply the creation of
new market facilitators (e. g., outgrower schemes), which in turns raises the need to
carefully look at potential agency problems in the design (are the stakeholders, more

specifically the smallholder farmers, in control and benefiting?)

163

Second, scale increase is needed throughout the chain. At the heart of the
dynamics created by supermarkets chains is an increase in scale, first at the retail level,
but then throughout the chain. Scale facilitates investment as well as the use of risk
reducing institutions such as contracts, standards and grades, and so on. This implies
group formation at the farmer level (investment capital, harvest schedules, volume), but
also (simultaneously) group formation at the wholesaler and retailer level (e.g., a new
type of smaller ‘farmer markets’ that can be located near residential areas and which
share a F FV procurement system). The size of group formation is determined by the
economies of size associated with transportation vehicles and irrigation systems as well
as by the volume requirements of the supermarkets (or other larger-scale buyers of F F V).

Third, one of the most critical challenges in any program designed to help
smallholder farmers in accessing modern supply chains is how to assist them in dealing
with the inherent higher working capital requirements. Smallholder growers are very
sensitive to working capital and cash flow issues as indicated by the frequently reported
break-down of contractual relationships between farmers and buyers because of the
former (either in temptation or in need) selling to brokers for cash (e.g., Jaffee and
Morton 1995). Therefore there is a strong need to develop and integrate creative solutions
that help keep working capital (and cash-outflows) down for farmers. The following are
some examples in this context: blended fertilizer, labor-driven pumps, factoring. One
project in Western Kenya involving 25 farmers has developed a fertilizer which is
blended specifically for a particular application in a particular area (based on a soil
analysis) and is marketing the fertilizer in small affordable bags (Okwemba 2004). Initial

results indicate yields that are 200% higher, which implies, inversely, that cash-strapped

164

farmers could reduce their expenditure on fertilizer without loss of yield. Kenya’s
MoneyMaker pumps, recently heralded as one of ‘Ten Inventions That Will Change the
World’ by Newsweek magazine (Stone 2003), provide a labor-intensive solution to the
required irrigation system. The pumps are (relatively) 10w cost in purchase ($60) and in
operation (using the labor of a worker pedaling the pumps’ pistons). Factoring, in Kenya
pioneered by the Kenya Gatsby Trust (an NGO), is a financial tool that could greatly help
farmers in dealing with the payment terms of supermarkets. Factoring entails buying out
(at a fee) the buyer’s payment obligation to the supplier after delivery took place. For
example, a farmer supplies Ksh5,000 worth of FFV to the supermarket and receives an
immediate payment of Ksh4,800 payment flom the factoring agent, rather than waiting
30 days for the payment to come flom the actual buyer. Factoring could be worked out
more systematically in order to take away the farmer’s temptation of defaulting on a

contract by selling to a broker because the latter offers cash on delivery.

165

BIBLIOGRPAHY

AC Nielsen (2002) Kenya Reported MBD ’s, AC Nielsen, Nairobi, Kenya.

Aflican Population and Health Research Centre (2003) Population and Health Dynamics
in Nairobi ’s Informal Settlements. Nairobi.

Akumu, W. (2003) ‘Supermarket Chain in KshlSm Loss’, Daily Nation, 3 March.

Al-Mazraooei, N., G.V. Chomo and A. Omezzine (2001) ‘Consumer Purchase Behavior
of Seafood Products in Oman.’ Journal of International Food and Agribusiness
Marketing, 13(4):5-22.

Barney, J .B. and W. Hesterly (1996) Organizational Economics: Understanding the
Relationship between Organizations and Economic Analysis, in SR. Clegg, C. Hardy
and W.R. Nord (eds.) Handbook of Organization Studies. Sage Publications:
Thousand Oaks

Battese G. E. (1992) ‘Frontier Production Functions and Technical Efficiency: A Survey
of Empirical Applications in Agricultural Economics.’ Agricultural Economics,

7: 1 85-208.

Bawden, R., Aust Stems, P., Harris, S. & Berdegue, J. (2002) Increasing Rural
Household Incomes in Kenya through Horticulture — A Design Proposal, Partnerships
for Food Industry Development - Fruits and Vegetables, Michigan State University.

Berdegue, J. A. (2001) Cooperating to Compete: Peasant Associative Business Firms in
Chile. Ph.D Thesis, Wageningen University and Research Centre, Department of Social
Sciences, Communication and Innovation Group. Wageningen, The Netherlands.

Berdegue, J .A., Balsevich, F., Flores, L. and T. Reardon (2004) ‘Central American
Superrnarkets’ Private Standards of Quality and Safety in Procurement of Fresh Fruits
and Vegetables’, Food Policy, forthcoming.

Bravo-Ureta, BE. and A.E. Pinheiro (1993) ‘Efficiency analysis of developing country
agriculture: a review of the flontier function literature.’ Agricultural and Research
Economic Review, 22:88-101.

166

Carter, MR. and K.D. Wiebe (1990) ‘Access to Capital and Its Impact on the Agrarian
Structure and Productivity in Kenya.’ American Journal of Agricultural Economics,
72(5):1146-1150.

CBS (2002a) Urban Household Budget Survey 1993/94. Nairobi: Central Bureau of
Statistics, Ministry of Finance and Planning.

CBS (2002b) Statistical Abstract 2002. Nairobi: Central Bureau of Statistics, Ministry of
Planning and National Development.

CBS (2002c) The New Kenya Consumer Price Index Users’ Guide. Nairobi: Central
Bureau of Statistics, Ministry of Finance and Planning.

CBS (2003) Statistical Abstract 2003. Nairobi: Central Bureau of Statistics, Ministry of
Planning and National Development.

Churchill, GA. (1999) Marketing Research: Methodological Foundations. Seventh
Edition. The Dryden Presstort Worth, TX, USA.

Daily Nation (2004) ‘Uchumi in an Upbeat Mood for Christmas’, 2 November.

Dijkstra, T. and TD. Magori (1992) Horticultural Production and Marketing in Kenya:
Part 2A: Horticultural Production in Nyandarua District, Report No. 47/1992, Food
and Nutrition Studies Programme, Afiican Studies Centre, Leiden, the Netherlands.

Dijkstra, T. (1997) ‘Commercial Horticulture by African Smallholders: A Success Story
flom the Highlands of Kenya’, Scandinavian Journal of Development Alternatives l6
(1): 49-74.

Dijkstra, T. (1999a) ‘Commercial Horticulture by Kenyan Smallholders’, in D.
Grossman, L. M. van den Berg and H. I. Ajaegbu (eds), Urban and Peri-Urban
Agriculture in Africa: Proceedings of a Workshop, Netanya, Israel, 23-7 June 1996.
Aldershot: Ashgate.

Dijkstra, T. (1999b) ‘Horticultural Marketing in Kenya: Why Potato Farmers Need
Collecting Wholesalers’, in H. L. van der Laan, T. Dijkstra and A. van Tilburg (eds),
Agricultural Marketing in Tropical Africa: Contributions fiom the Netherlands.
Aldershot: Ashgate.

Dijkstra, T. (2001) ‘Applying Marketing Channel Theory to Food Marketing in
Developing Countries: Vertical Disintegration Model for Horticultural Marketing
Channels in Kenya’, Agribusiness l7 (2): 227-41.

167

Dolan, C. S. and Humphrey, J. (2000) ‘Governance and Trade in Fresh Vegetables:
Impact of UK Supermarkets on the African Horticulture Industry’, Journal of
Development Studies 37 (2): 147-77.

Dolan, CS. (2001) ‘The Good Wife: Struggles over Resources in the Kenyan
Horticulture Sector.’ Journal of Development Studies, 37(3):39-70.

Dries, L., Reardon, T. and Swinnen, J. (2004) ‘The Rapid Rise of Supermarkets in
Central and Eastern Europe: Implications for the Agrifood Sector and Rural
Development’, Development Policy Review 22 (5): 525-56.

East Aflica Seed Co. (2002) Grower ’s Guide. Nairobi, Kenya.

Econstats (2004) Available at http://www.econstats.com/, accessed May 2004.

FAO (2004) Available at http://famt.fao.org/faostat/collections?subset=2_1griculture.
accessed May 2004.

F eder, G., Just, RE. and D. Zilberrnan (1985) ‘Adoption of Agricultural Innovations in
Developing Countries: A Survey.’ Economic Development and Cultural Change,
33(2): 255-98.

F eeny, A., Vongpatanasin, T. and A. Soonsatharn (1996) Retailing in Thailand.
International Journal of Retail and Distribution Management, 24(8):38-45.

Ghezan, G., Mateos, M. and Viteri, L. (2002) ‘Impact of Supermarkets and F ast-Food
Chains on Horticulture Supply Chains in Argentina’, Development Policy Review 20
(4): 389-408.

GOK (Government of Kenya). (1994) Development Plan 1994-1996,
NairobizGovemment Printer.

Greene, W.H. (2000) Econometric Analysis, Upper Saddle RiverzPrentice-Hall, pp 1004.

Heckman, J .J . (1979) ‘Sample Selection Bias as a Specification Error.’ Econometrica,
47(1):153-162.

168

Heston, A., Summers, R. and A. Bettina (2002) Penn World Table, Version 6.1, Center
for International Comparisons at the University of Pennsylvania (CICUP) (available
at http://pwt.econ.upenn.edu/php site/pwt61 forrnphp 24, accessed 24 May 2004).

Hu, D., Reardon, T., Rozelle, S., Tirnmer. P. and H. Wang (2004) ‘The Emergence of
Supermarkets with Chinese Characteristics: Challenges and Opportunities for China’s
Agricultural Development.’ Development Policy Review, 22(5):557-586/

Hueth, B., Ligon, E., Wolf, S. and Wu, S. (1999) ‘Incentive Instruments in Fruits and
Vegetables Contracts: Input Control, Monitoring, Measurements, and Price Risk’,
Review of Agricultural Economics, 21 (2):374-89.

Jaffee, S. (1994) ‘Contract Farming in the Shadow of Competitive Markets: The
Experience of Kenyan Horticulture’, in P. Little and M. Watts (eds), Living under
Contract: Contract Farming and Agrarian Transformation in Sub-Saharan Afi'ica.
Madison, WI: University of Wisconsin Press.

Jaffee, S. (1995) The Many Faces of Success: the Development of Kenya’s Horticultural
Exports’ in J affee, S. and J. Morton (eds) Marketing Africa ’s High- Value Foods:
Comparative Experiences of an Emergent Private Sector. World Bank:Washington,

D.C.

Jaffee, S. and J. Morton (1995) Marketing Africa ’s High- Value Foods: Comparative
Experiences of an Emergent Private Sector. World Bank:Washington, D.C.

Karnau, M. (2000) The Way Forward in Export Oriented Smallholder Horticulture.
Background paper for a Stakeholder Consultation Meeting, TEGEMEO Institute of
Egerton University, Nairobi, 8 February.

Kaplinsky, R. and M. Morris (2001) A Handbook for Value Chain Research. Report for
the IDRC.

Ministry of Agriculture (2003) Farm Management Guidelines 2003, Limuru Division,
Ministry of Agriculture.

Ministry of Agriculture (2004) Farm Management Guidelines 2004, Lari Division,
Ministry of Agriculture.

169

MoARD (2000) Local and Export Vegetables Growing Manual. Nairobi: Agricultural
Information Resource Centre, Ministry of Agriculture and Rural Development.

MoARD (2002) Horticulture Division Annual Report 2002. Nairobi: Horticulture
Division, Department of Agriculture and Livestock Production, Ministry of
Agriculture and Rural Development.

Mungai, J ., Ouko, J. & Heiden. (2000) Processing of fruits and vegetables in Kenya.

Obudho, R.A. (1997) ‘Nairobi: National Capital and Regional Hub’, in C. Rakodi (ed)
The Urban Challenge in Africa: Growth of its Large Cities. Tokyo: UN University
Press. (available at http://www.unu.edu/unupress/unupbool§/uu26ue/uu26ue00.htrn.
accessed 4 May 2004).

Okwemba, A. (2004) ‘Sh4 Fertilizer Makes All the Difference.’ Daily Nation, 4
November.

Oregon State University (2002) Commercial Vegetables Production Guides. Collards
and Kale. Accessed online at oregonstate. edu/Dept/NWREC/collards. htrnl, 20
September 2004.

Peterson, H.C., Wysocki, A. and SB. Harsh (2001). ‘Strategic Choice Along the Vertical
Coordination Continuum.’ International Food and Agribusiness Management
Review, 4: 149-166.

Porter. ME. (1980) Competitive Strategy. Free PresszNew York.

Pralahad, C.K. & Hammond, A. (2002) Serving the World’s Poor, Profitably. Harvard
Business Review, September: 48—57.

Reardon, T. and CB. Barrett (2000) Agroindustrialization, globalization and international
development: An Overview of Issues, Patterns and Determinants. Agricultural
Economics, 23: 195-205.

170

Reardon, T. and Berdegue, J .A. (2002) ‘The Rapid Rise of Supermarkets in Latin
America: Challenges and Opportunities for Development’, Development Policy
Review 20 (4): 317-34.

Reardon, T., Timmer, C.P., Barrett, CB. and Berdegue, J. (2003) ‘The Rise of
Supermarkets in Aflica, Asia and Latin America’, American Journal of Agricultural
Economics 85 (5): 1140-6.

Reardon, T. (2004) Supermarkets and Agricultural Development in Indonesia: First
Impressions. Report for USAID via the RAISE/FPSA project, February.

Rodriguez, E., M. Berges, K. Casellas, R. Di Paola, B. Lupin, L. Garrido and N. Gentile.
(2002) Consumer Behavior and Supermarkets in Argentina. Development Policy
Review, 20(4): 429-439.

Sadoulet, E. and A. de Janvry (1995) Quantitative Development Policy Analysis. The
John Hopkins University PresszBaltimore.

Shah, H. T. (2004) Kenya: Exporting Out of Afiica: Kenya’s Horticulture Success Story.
Paper presented at Scaling up Poverty Reduction: A Global Process Conference,
World Bank, Shanghai, 25-7 May.

Smits, J. (2003) Estimating the Heckman Two-Step Procedure to Control for Selection
Bias with SPSS. Available at homeplanet.nl/~smits.ieroen/selbia_§/selbia_s.html.
accessed November 23, 2004.

Stone, B. (2003) ‘Waterworks’. Newsweek, June 30 — July7.

Uchumi (2003) Annual Report and Accounts 2003. Nairobi: Uchumi.

UN Habitat (2004) Available at www.unhabitat.orglhabrdd/conditions/eafrica/kenvahtm,
accessed 3 May 2004.

UN Population Division (2003) World Population Prospects: The 2002 Revision
Population Database. Available at http://esz_1.un.orglunpp/kaOdata.asp, accessed 3
May 2004.

171

UN Statistics Division (2003) Available at hfip://unstats.un.org[unsd/cdbdemo/. accessed
4 May 2004.

Wachira, N. (2002) ‘Uchumi Bets Big on Specialty Shops’, Financial Standard, 25
November-2 December.

Wachira, N. (2004) ‘DJ CK Calls it a Day at Uchumi’, Daily Nation, 20 July.

Wahome, M. (2001) ‘Giant Supermarkets in Battle for Supremacy’, Daily Nation, 11
December.

Weatherspoon, D. D. and Membreflo, T. (2004) PFID-F62 V: Facilitating Market
Linkages for Small-Scale Farmers in Nicaragua. PFID Project Brief presented at the
Supermarkets and Agricultural Development Conference, Washington, DC, 28 April.

Weatherspoon, DD. and Reardon, T. (2003) The Rise of Supermarkets in Afiica:
implications for agrifood systems and the rural poor. Development Policy Review,
21(3): 333-355.

Williamson, QB. (1985) The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting. Free PresszNew York.

Williamson, CE. (1991) ‘Strategizing, Economizing, and Economic Organization.’
Strategic Management Journal, 12(Special Winter):75-94.

Wooldridge, J .M. (2000) Introductory Econometrics: A Modern Approach. South-
Westem College Publishing, 824p.

World Bank (2004) 2003 World Development Indicators. Washington DC: The World
Bank.

World Gazetteer (2004) Available at www.world-gazetteer.com/t/t_ke.htrn, accessed 4
May 2004.

Zhang, X. (2002) The Dynamics of Chinese Consumers: The Case of Shanghai Food
Consumption. Journal of International Food and Agribusiness Marketing, 14(1):47-66

172

   

1lllllillllflllllllllfllflllIllllllu