3 4:.

u \!!A
«I {a .
.1... . .43.
.33 .u. . .
ll! -
4": u- ‘. ‘
.‘tIAAv

..

n4.

en. a...
t limp}:
nan ... Jr“

1..

".342.
2:2; ..
g. x.
.
a1. . w.
: inkwwmn.

[A .I V I
(I; .19:
v.3 . .. .. .

:7

§£§® _

V:

“m.n.me.mm$m,w , . ‘
. 4 v I

 

h ”2 _ LIBRARY
‘00 ’ Michigan State
University

This is to certify that the
dissertation entitled

ESSAYS ON PRODUCER’S PARTICIPATION, ACCESS AND
RESPONSE TO THE CHANGING NATURE OF DYNAMIC
DOMESTIC MARKETS IN NICARAGUA AND COSTA RICA

presented by

FERNANDO BALSEVICH

has been accepted towards fulfillment
of the requirements for the

degree in gAricultural Economics

 

 

mméa/M

Major Professor’ 5 Signature

Decerr/7-8i’ ,2; 2,006

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:IClRC/DaIeDue,indd-p.1

ESSAYS ON PRODUCERS’ PARTICIPATION IN, ACCESS TO, AND
RESPONSE TO THE CHANGING NATURE OF DYNAMIC DOMESTIC
MARKETS IN NICARAGUA AND COSTA RICA

By

Fernando Balsevich

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Agricultural Economics

2006

ABSTRACT

ESSAYS ON PRODUCERS’ PARTICIPATION IN, ACCESS TO, AND
RESPONSE TO THE CHANGING NATURE OF DYNAMIC DOMESTIC
MARKETS IN NICARAGUA AND COSTA RICA

By

Fernando Balsevich

This dissertation analyses the relations among tomato and cattle producers and
their market channels in Nicaragua and Costa Rica. The first essay analyses the
relations among 145 tomato producers and three market channels in Nicaragua. It
analyzes the recent development of supermarket chains, their changes in
procurement systems and the implications for producers. Then it explores market
channel impact on producers’ profitability and technology. Results indicate that
producers selling to the supermarket chains generate higher profits than producers
selling to traditional markets because they are more productive. Producers selling
to supermarket chains intensify the use of inputs. However, producers selling to
the leading and second-tier supermarket chains use a labor-biased and capital-
biased technology, respectively, as compared to the traditional channel.

The second essay focuses on the determinants and effects of the participation
of cattle producers in the supermarket channel, export processor channel, and

traditional auction channel. It begins with the analysis of the market channels

using qualitative data from 50 interviews of retailers, processors, auction market
managers, and other key informants in Costa Rica and Nicaragua, two widely
differing cases. It then analyzes patterns and supplies of producers by channel
using farm level data (from the authors’ survey of farmers in 2004) from 300

farms in the two countries.

This work is dedicated to Fernando Enrique, my family and friends.

iv

ACKNOWLEDGMENTS '

I thank my major professor and dissertation adviser Dr. Thomas Reardon for his
support and constructive comments. I also Dr. Robert Myers, Dr. Christopher Peterson,
Dr. Dave Weatherspoon, and Dr. Lawrence Busch for their kindness, support and
participation on my research committee. I also benefited from comments and support
from my graduate students fellows, friends, and PFID-F&V team members.

I’m thankful to comments from Julio A. Berdegué from Rimisp — Latin American
Center for Rural Development, Santiago, Chile and comments from Paul Schuetz and
Edwin Perez from the International Livestock Research Institute (ILRI).

The dissertation has been part of a coordinated research program in Nicaragua,
Costa Rica, and Guatemala in 2004, with the participation of the following projects and
donors, in no particular order: (1) “Assistance for Trade Capacity Building in Relation to
the Application of Sanitary and Phytosanitary Measures”(RAISE-SPS), funded by
USAID, primed by DAI and implemented for this project by Michigan State University;
(2) The “Central America Beef Project”, of the International Livestock Institute (ILRI),
funded by the Common Fund for Commodities; (3) “Regoveming Markets, Phase 1”, a
project coordinated by the International Institute for Environment and Development
(IIED), the Royal Tropical Institute (KIT), and Rimisp-Centro Latinoarnericano para el

Desarrollo Rural, and funded by the Department for International Development (DFID)

of the United Kingdom; (4) “Partnerships for Food Industry Development — Fruits and
Vegetables” (PFID-FF V), funded by USAID, and implemented by Michigan State
University; and (5) The Agribusiness Development Project of the Centro Intemacional de

Agricultura Tropical (CIAT).

vi

TABLE OF CONTENTS

LIST OF TABLES ........................................ viii
CHAPTER 1: INTRODUCTION ....................................................................................... 1
CHAPTER 2: SUPERMARKETS, NEW-GENERATION WHOLESALERS, TOMATO

FARMERS, AND NGO’S IN NICARAGUA .................................................................... 4
1. INTRODUCTION ...................................................................................................... 5

2. THE THREE SUPPLY CHAINS ............................................................................... 8
(a) Supply Chain Analysis: The Vertical Coordination Continuum ........................... 8

(b) The Three Tomato Supply Chains’ Structures .................................................... 12

3. PRODUCERS IN THE THREE MARKET CHANNELS ....................................... 23
(a) Data and sample .................................................................................................. 25

(b) Farm Characteristics, Technology Use, and Profits ............................................ 26

(c) The Determinants of Producer Channel Choice .................................................. 32

(d) Comparison over Channel Strata of Technology and Efficiency ........................ 37

(e) Output Supply and Input demand functions ........................................................ 41

4. CONCLUSIONS AND IMPLICATIONS ................................................................ 45

CHAPTER 3: ESSAY ON CATTLE PRODUCERS’ PARTICIPATION IN MARKET
CHANNELS IN CENTRAL AMERICA: SUPERMARKETS, PROCESSORS, AND

AUCTIONS ...................................................................................................................... 59
1. INTRODUCTION ....................... 6O
2. CATTLE SUPPLY CHAIN COORDINATION MECHANISMS .......................... 63
(a) Data for the Qualitative Analysis of Supply Chains ........................................... 63
(b) Conceptual Framework for Supply Chain Analysis: The Vertical Coordination
Continuum ................................................................................................................. 64
(0) Beef supply chains (market channels) in Nicaragua ........................................... 68
(d) Beef supply chains (market channels) in Costa Rica .......................................... 75
(e) Linking Descriptive Analysis with a Vertical Coordination Model .................... 82
3. PRODUCER CHARACTERISTICS BY MARKET CHANNEL ........................... 84
4. ECONOMETRIC MODELS TO EXPLAIN MARKET CHANNEL
PARTICIPATION AND STRATA SUPPLY DIFFERENCES ................................... 90
(a) Random Utility Model to Determine Producer Channel Choice ......................... 91
(b) Market Channel Output Supply functions ........................................................... 98
5. CONCLUSIONS AND IMPLICATIONS .............................................................. 103
CHAPTER 4: CONCLUSIONS ..................................................................................... 1 16
REFERENCES .................................................................. ' ............................................. 120

vii

LIST OF TABLES

Table 1: Vertical Coordination Strategies in the Three Tomato Supply Chains .............. 47
Table 2: Farm Characteristics ......................................... 48
Table 2: Farm Characteristics (Continuation) ................................................................... 49
Table 3: Tomato Production Characteristics ..................................................................... 50
Table 3: Tomato Production Characteristics (Continuation) ............................................ 51
Table 4: Tomato Production Characteristics ..................................................................... 52
Table 4: Tomato Production Characteristics (Continuation) ............................................ 53
Table 5: Marketing Channel Selection Estimation Results (WESML) ............................ 54
Table 6: Second stage: Production Function Estimation Results ..................................... 55

Table 7: Marginal Product Value (MPV) and Average Product Value (APV) by
Marketing Chain (calculated from second stage Cobb Douglas production function).

................................................................................................................................... 56
Table 8: Output Supply and Intermediate Input demand Function Estimation Results
(Zellners) ................................................................................................................... 57
Table 8: Output Supply and Intermediate Input demand Function Estimation Results
(Continuation) ........................................................................................................... 58
Table 9: Vertical Coordination Strategies in the Beef Supply Chains ............................ 107

Table 9: Vertical Coordination Strategies in the Beef Supply Chains (Continuation) 108

Table 10: Farm and Producer Characteristics ................................................................. 109
Table 10: Farm and Producer Characteristics (Continuation) ........................................ 110
Table 11: Producers' sales and purchases of cattle by channel ....................................... 111
Table 11: Producers' sales and purchases of cattle by channel (Continuation) .............. 112
Table 12: Producers' market channel determinants ........................................................ 113
Table 13: Producers' supplies by market channel ........................................................... 114

viii

Table 13: Producers' supplies by market channel (Continuation) .................................. 115

ix

CHAPTER 1: INTRODUCTION

There have been a number of contributions in understanding the complexities of
the factors of moving producers from a subsistence orientation to commercial production.
Although no ambiguity seems to exist for rural developers in the importance of
understanding the markets for rural products, Pingali & Rosegrant (1995) have pointed
out the challenges faced in smoothing the transition from subsistence to commercial

production systems.

Two sets of research studies seem to contribute to understanding the rural markets.
One set relates to producer’s market participation studies such as Goetz (1992), Key et a1.
(2000), Bellemare and Barrett (2004), and Holloway et a1. (2004). Another set of studies
focuses on multi-market factors as market channel determinants. Examples of the latter
are Hobbs (1997) which compares the determinants of auction versus processor market
channels; Ruben (2003) compares the determinants of local versus wholesale market

channels.

Whether the research focus is to explore market participation determinants or
multi-market determinants, the objective is to understand the factors that affect
producer’s access to the market or their access to a market channel. Better understanding
the determinants can contribute to the actions and efforts of those actors that aim to
implement market-led development strategies. It can be that the final outcome is to
generate employment, increase income, reduce poverty reduction, and/or increase

supplies for healthier consumer nutrition.

Rural markets are always changing and the research focus ought to keep pace in
order to find better and more tailored ways to implement reasoned development
interventions. An example of changes in the market that this dissertation aims to explore

is the supermarket’s increasing role in marketing rural products.

Note that as supermarkets in Latin America and the Caribbean(LA) moved from
10-20% participation in the urban food retail sector in 1990 to 50-60% in 2000 (Reardon
and Berdegué, 2002), changes are occurring much faster than in the US and Europe
(Berdegue’ et. al, 2005). A number of studies in LAC have been done that provide
country-specific evidence of the rise of supermarkets such as Ghezan et a1. (2002) and
Gutman (2002) in Argentina; Farina (2002) in Brazil; Faiguenbaum et al. (2002) in Chile;
Alvarado and Charmel (2002) in Costa Rica; and Schwentesius and Gomez (2002) in

Mexico.

Broadly, new buyers are synonymous with new market channels for rural
products. A key issue is that the literature points out that the supermarkets have stricter
product and transaction requirements as compared to traditional retailers. What does this
mean for rural products market access, for those aiming at increasing market access to
reduce poverty levels? How will these new actors in the market impact the rural

producer’s markets? The research gap is twofold:

(1) Lack of inclusion of supermarket channels in studies of the determinants and effects
of producers’ market access; (2) Lack of empirical evidence in supermarket studies

regarding impacts on producers.

The main objective of this dissertation is to contribute to cover the twofold gap

mentioned in the last paragraph. Despite the increasing importance of supermarkets in

Latin American countries, however, there has been little empirical research on three
issues which form the focus of this dissertation: (1) how do cattle and horticultural supply
chains differ over market channels (supermarket chains, auction markets, processors and
traditional markets) and what implications do these differences have for the marketing
environment facing producers? (2) what are the determinants of channel (supply chain)
choices of farmers; (3) what are the correlates of market channel choices and producer’s

technologies and net incomes?

The dissertation is composed of two essays. In the first essay, three market
channels (with different stage of development) of tomatoes in Nicaragua were mapped
backwards to their suppliers. Several in-depth interviews were conducted in order to
understand the recent development of supermarket chains, and their changes in
procurement systems. Statistical and econometric data analysis from a sample of 145
producers selling to those market channels is presented in order to draw conclusions and

implications for producers.

In the second essay, six market channels of cattle producers (three in Nicaragua
and three in Costa Rica) are analyzed. In-depth interviews were conducted in order to
understand the recent development of supermarket chains, industry changes and in the
case of Costa Rica, development of auction markets. Statistical and econometric data
analysis from a sample of 300 producers is presented. The analysis focuses on the
determinants and effects of the participation of cattle producers in the supermarket

channel, export processor channel, and traditional auction channel.

CHAPTER 2: SUPERMARKETS, N EW-GENERATION WHOLESALERS,

TOMATO FARMERS, AND NGO’S IN NICARAGUA

Abstract
Based on a survey of 145 tomato farmers, and interviews with supermarket chains,
NGOs, wholesalers, and farmer organizations in 2004, this paper examines the
determinants and effects of farmers’ participation in supermarket channels, with and
without assistance from NGOs in “business linkage” programs. It finds that absent that
assistance, the farmers that work with supermarket chains tend to be the “upper tier” of
small farmers, better capitalized with various assets. The smaller and less-capitalized
farmers that work with supermarkets tend to do so in association with NGO assistance.
Despite higher input expenditures and entry requirements, farmers in the supermarket
chain earn more per hectare. The paper ends by raising the issue of whether this
development program approach can be scaled up, and wrestles with the tradeoff of
helping poor farmers gain access to dynamic markets, of making it affordable at a larger
scale by national governments with tight budgets, and of undertaking field programs that

are market-sustainable and market-responsive.

Supermarkets, New-Generation Wholesalers, Tomato Farmers, and
NGOs in Nicaragua

1. INTRODUCTION
Driven by rising incomes and urbanization, as well as foreign investment and

procurement technology changes, the share of supermarkets in food retail in Latin
America rose from a mere 10-20% in 1990 to 50-60% by the early 20005, displacing
small shops and open-air markets (Reardon and Berdegue', 2002). That trend started
somewhat later in Central America and has gone less far, reaching 20-40% (depending on
the country) of food retail by the early 20008, and still rising. One of the poorest countries
in the region (circa 400 dollars per person GDP/capita) and a small country (four million),
Nicaragua, has shared that trend, with the share reaching 15-20% today, and rising

(Berdegué et al., 2005).

Food retailing in general, and tomato retailing in particular, have, until very
recently, been undertaken exclusively in the “traditional retail sector” in Nicaragua.
That sector is composed of many small grocery stores, local wet—markets, street
vendors and retail sections of large central markets. These traditional retailers procure
their produce directly from farms (when the retailers are in small towns and villages)
and from the traditional wholesale markets in the main and secondary cities (Leiva,

2004)

A new set of actors in food retail in Nicaragua is at an early stage of emergence
— supermarkets. In the early 19903, supermarkets had the tiniest of niches, with only 5%

of food retail; that has increased to 15 to 20% in 2003, still an early stage but the

beginning of a factor affecting food markets (Berdegué et al. 2005). The supermarket
sector “took off” afier the end of the revolutionary period and the liberalization of
foreign direct investment (FDI). As is typical for supermarkets, their penetration of the
overall food market exceeds that of their share in fresh produce retail, where they now
have only about a 10% share. It is thus particularly interesting to understand the effect
of the new retail channels on food markets and farmers at this early stage, for its own

sake, and to form a base for observation as it grows over time.

Today the Nicaraguan supermarket sector is made up of two supermarket chains
(the leader, CSU, and the second-tier chain, La Colonia) (with 37 stores) with about two-
thirds of the market, and 23 independent supermarkets. The two chains are as follows. (1)
CSU is Corporacio'n de Supermercados Unidos, financed with Costa Rican capital,
which entered Nicaragua in 1994 with 3 stores. By 2005 CSU had 30 stores targeting
higher-income consumers under the label “La Union” and lower-income consumers
under the label “Pali.” Pali stores are 80% of the chain, and the fastest growing segment
as CSU expands into lower-middle and working poor consumer markets and into
secondary and tertiary cities in the provinces. CSU is part of the Central American Retail
Holding Company (CARHCO), which through August 2005 (hence during our study
period) was a joint venture among CSU, La Fragua (the leading chain in Guatemala) and
Ahold (the third largest food retail chain in the world, based in the Netherlands), with 1.5
billion US dollars in sales and 363 stores in Guatemala, El Salvador, Honduras, Costa
Rica, and Nicaragua.1 (2) “La Colonia” is the largest locally owned chain. This chain
grew much more slowly than CSU (the difference between a large chain in an established

market vigorously investing, versus a local chain growing from the local base). It started

about the same time as CSU, went from three stores in 1990 to seven stores in 2004, and
has focused only on Managua, the capital city.

The rapid rise of supermarkets, and hence changes downstream in the agrifood
system, can be hypothesized to be changing market conditions facing farmers. Compared
to traditional retailers, supermarkets generally have different and more demanding
product and transaction requirements. Despite the increasing importance of supermarkets
in Latin American countries, however, there has been little empirical research on three
issues which form the focus of this paper: (1) how do horticultural supply chains differ
over types of supermarket chains (leading versus second-tier) and between supermarkets
and traditional markets, and what implications do these differences have for the
marketing environment facing producers? (2) what are the determinants of channel
(supply chain) choices of farmers (between supermarket and traditional market channels);
(3) what are the correlates of market channel choices and producers’ technologies and net
incomes? This paper focuses on these issues, within the context of tomato growers and
supermarkets in Nicaragua. The analysis is based on field interviews with supermarkets,
wholesalers, and NGOs, and a representative survey of farmers.

The paper’s structure follows the structure of the research questions and
proceeds as follows. Section 2 focuses on the tomato supply chains in different market
channels: the two supply chains to supermarkets (the leading supermarket chain, CSU,
and that the second supermarket chain, La Colonia) and to the traditional market (the
main public wholesale market). The section starts with a conceptual framework used to
analyze the supply chains. Definitions of five levels in the vertical coordination

continuum are presented - from low to high levels of control. Next we present the

method used to collect our data and describe characteristics of the three tomato supply
chains. Section 3 focuses on the growers participating in the different market channels.
We start by describing the characteristics (assets, technologies, net incomes) of the
growers. We then present results of regressions of the determinants of market channel
choice by producers and, with endogenous stratification by channel, the output supply,
input demand, and technology choices made by farmers. The last section provides

conclusions and discusses implications for policy.

2. THE THREE SUPPLY CHAINS

In this section, we briefly analyze the three supply chains of tomatoes (to the
leading supermarket chain, to the second-tier chain, and to the traditional market). Note
that we use several terms interchangeably: “procurement systems,” “supply chains,” and
“market channels” — as these represent the same thing viewed from different perspectives
- the retailer, the system analyst, and the farmer. We first present a conceptual
framework that we use to analyze the nature of each supply chain’s coordination
mechanisms, of interest in itself and because those mechanisms in turn have an effect on
the market channel conditions facing the farmers. Second, we present the three supply
chains’ structures, themselves outcomes of strategic coordination choices by the retailers

and wholesalers, and describe changes over time that have occurred in these mechanisms.

(a) Supply Chain Analysis: The Vertical Coordination Continuum

Our conceptual model is based on the work of Peterson, Wysocki, and Harsh

(2001), hereafter the PWH model. The model aims to be a decision making tool that

helps to understand how and why the agents along the supply chains (supermarkets,
wholesalers, producers) make strategic choices from alternatives ranged along a vertical

coordination continuum.

Vertical coordination can be defined as the configuration and control of the stages
of production and marketing systems — including the specifications asked of the producer
(such minimum size and color of the fruit) by which the tomatoes are produced, who is
involved (such as whether farmer delivers direct to the supermarket or via an
intermediary), and under what marketing conditions (such as payment terms) until the
tomatoes reach the final consumer. As there are potentially many stages and agents
involved in production and marketing of a product such as tomatoes, there are also many
relations under which the transactions take place. Characterizing vertical coordination
mechanisms thus also involves identification of the institutional arrangements (what they
are, such as whether a spot or contract relation, who defines them, and the incentives put

in place to implement them, such as price premia) governing those relations.

The continuum refers to the set of choices that a given agent in a supply chain has
to vertically coordinate its procurement system (for example, procuring tomatoes). As
there may be many ways to vertically coordinate the procurement of tomatoes, some
criteria to categorize the alternatives may be useful. The PWH model defines the vertical
coordination continuum as an array of five groups of strategies: spot markets,
specification contracts, relation-based alliances, equity-based alliances, and vertical
integration — moving from the least intensity to the greatest intensity of vertical

coordination.

As coordination intensity increases, so does the nature of monitoring and control.
(1) The least intensive coordination is the spot market, where the coordination
mechanism is based mainly on price discovery and the decision of whether to engage in
the transaction, and there is no coordination mechanism beyOnd the transaction. This
typifies an auction market. (2) In the specification-contract coordination mechanism,
agents agree to a legal contract, the specifications of which set incentives and
deliverables, as well as use of third parties to enforce the contract. (3) The relation-based
alliance establishes a relation between the agents in which benefits and risk are shared
and mutual objectives identified. Parties collaborate in conflict resolution and monitoring.
(4) The equity-based alliance involves some financial commitments between the parties
(such as equity capital investment in a joint venture), and the formation of a third entity
distinct from the parties that has a formal organizational structure. (5) The most intensive
coordination is vertical integration, where the two parties become one organization (by a
merger or acquisition) which has full control over the transaction, with clear rules and

centralized control.

There are several salient determinants of the choice by supply chain actors of the

coordination intensity on the above spectrum.

First, agents’ coordination choices depend on costs incurred from coordination
errors. Agents tend to want more intense vertical coordination the more if coordination
errors (for example, quality or volume supplied is inadequate) are costly (for example in
foregone sales). Coordination errors can occur because of intentional opportunism of
economic actors or unintentionally due to bounded rationality of estimations in

production plans (Williamson, 1973).

10

The cost of coordination errors is a function of two things. (1) “Asset specificity”
refers to the extent to which an asset used in the activities of a given actor, is specific to
meeting the requirements of the buyer downstream in the chain. If a producer uses
packing equipment that is designed for the specifications of a given buyer (say a
supermarket chain), and difficult to use to pack for other clients, that asset is specific, and
coordination errors harming access to the client means under use or non-use of a costly
asset. Or a producer may have to enter a relation with a wholesaler who has a packing
plant with the needed equipment to pack for that supermarket chain, and coordination
errors between the farmer and the wholesaler then block the farmer from access to the
chain. (2) “Complementarity” refers to complementary actions by several actors in the
supply chain, which then require coordination for any given action to be profitable. For
example, if a supermarket chain starts an “organic vegetables category” in its produce
section, it requires investments by farmers in organic practices, and vice versa. Over-
capacity in organic production without a downstream agent to sell to the niche market
segment who buys organic produce, or over-capacity at the marketing end, both spell

coordination errors.

Note that the costliness and likelihood of coordination errors may differ over
segments of the supply chain, and thus within a supply chain one may observe different
coordination intensities/mechanisms. For example, in a tomato supply chain where there
are three agents (supermarket, wholesaler, and producer) and two relations (supermarket-
wholesaler and wholesaler-farmer). The supermarket and wholesaler may choose a

relation-based alliance, while the wholesaler and farmer might relate in a spot market.

11

Second, agents’ coordination choices depend on the cost of implementing a given
coordination strategy. These costs are varied and a function of the product, the market,
and the relation; an example is contract monitoring capacity investments (such as in
agronomists to check producer practices or an on-site lab to Check pesticide residues on

fruit received), or skill acquisition to meet production or post-harvest standards.

Finally, the agents weigh the above two sets of costs (costs of coordination errors
and costs of strategy implementation), against the returns(and the distribution of returns,
hence the risks) of one or another strategy. Unless they have experience with both, or try
both simultaneously, they can only estimate the likelihoods of such events taking place,
given the (always limited) information they have about the other participants and other

exogenous factors.

(b) The Three Tomato Supply Chains’ Structures

For the analysis of the three tomato supply chains, we undertook a rapid
appraisal of the market channels in a field study in July/August 2004. This involved.
thirty semi-structured interviews with the two supermarket chains, several traditional
retailers, wholesalers serving the supermarket chains and several serving only the
traditional market, NGOs providing technical and other assistance to growers, and
producer groups. Besides dwelling on the basic information concerning transactions
(who, what, where, when), the interviews with retailers and wholesalers treated in
depth the “how” questions bearing on the coordination mechanism: (1) the use (or not)
of contracts, verbal or written, formal or informal? And the history of the
coordination mechanisms used; (2) the terms of the contracts, in terms of prices,

attributes required (such as quality standards, delivery times and volumes); (3) the

12

reasons for the agreements, in terms of the kinds of costs and benefits discussed
above for alternative coordination mechanisms versus the one chosen; (4) the way the
contracts are implemented (who designs, who implements, how products are selected,
how prices are negotiated, how delivery and payment take place, and so on); (5) how

the agreements are enforced.

Table 1 summarizes the structures of the three supply chains, including for
each segment (defined as a transaction, say producer with wholesaler, or wholesaler
with retailer) of the three supply chains the coordination mechanism(s). The

following discussion follows the structure and points of Table 1.

Leading Supermarket Chain ’s (CSU) tomato supply chain. CSU sources all its
produce from Hortifruti, a specialized, dedicated wholesaler. Hortifruti is also itself a
regional multinational operating in the same countries as CSU (Costa Rica, Nicaragua,
and Honduras) and is in the same Costa Rican holding company as CSU. Hortifruti is a
new-generation wholesaler that started its local Nicaraguan subsidiary operations in the
mid 19903. We call Hortifruti and its ilk “new-generation wholesalers” because of their
differences from traditional wholesalers in their contractual or semi-contractual relations
with clients, dedication to the supermarket sector (and other modern food industry clients
such as fast food chains), and tendency to go beyond mere spot market relations with
farmers. The new-generation wholesaler procures, selects and sorts, and packs for the
supermarkets, and delivers to the recently started distribution centers of the chain. Often

they also stock supermarket shelves as well.

13

The coordination mechanism used between CSU and Hortifruti corresponds to a
hybrid of the “equity-based alliance” and “vertical integration” mechanisms discussed
above. Full vertical integration requires centralization of operations between the
companies, which is not the case between CSU and Hortifruti. Full equity-based
alliance implies a third entity to manage transactions between the companies — again,
not the case between CSU and Hortifruti, who deal directly. The relation differs from
the other supply chain relations we study mainly because it involves equity linkages

between the parties.

Initially, from its entry in Nicaragua in 1994 to the coming of Hortifruti in 1998,
CSU sourced directly from traditional wholesalers (in the wholesale market). Moreover,
even Hortifruti, once it became the source for CSU’s produce, also relied heavily on the
wholesale market for several years, before establishing direct relations with producers
in the early 20005. There were thus two consecutive shifts, for common reasons relating
to the need to reduce coordination costs: (1) CSU’s shift from wholesale market to
Hortifruti, and (2) Hortifruti’s shift from sourcing fiom the wholesale market (as a
temporary strategy as it set up a sourcing system similar to what it already had in Costa

Rica) to sourcing mainly directly from farmers.

What initially drove the choice of CSU (and initially, Hortifruti) to source from
the traditional wholesale market? The latter sold and sells large volumes, aggregating
from over thousands of tiny farms across Nicaragua, draws from across regions so there
was at least a small degree of “season stretching”, and sells the gamut of produce

grades.

14

However, the main problem with the traditional wholesale market, leading to
the main coordination error, is that there is extremely sharp seasonality of availability
of a given product, relative to the constant need of the retailer. This relates to what we
can term farmers’ and traditional wholesalers’ “supply—periOd asset specificity”.
Moreover, even when product was available, the quality (commercial grade based on
appearance such as size, blemishes, and color) was perceived as highly inconsistent
over traders and over time within the market. Working with many small brokers also
led the retailer to a perception of what we translate as congestion costs. CSU and then
Hortifruti felt that the traditional brokers did not perceive a need or incentive to
cooperate in meeting the quality specifications, which requires classification. Hortifruti
noted that around 2000, when they shifted toward sourcing from farmers, there were

significant gains in cutting out the market middleman in terms of costs.

Around 2000, faced with the limitations of the wholesale market, CSU turned to
Hortifruti and Hortifruti initiated new practices and made investments, and turned to
direct relations with farmers. The changes made were as follows, implemented

progressively over the past several years.

First, CSU and Hortifruti agreed to a set of private quality standards and
transaction attributes, specified in writing in manuals that are a de facto contract
between them. The standards relate to produce quality (minimum size, form, ripeness,
insect and mechanical damage, color, and firmness) and transaction requirements
(packing, volumes, and consistency). Those private standards were brought from their
Costa Rican headquarters, adapted to the Nicaraguan consumers, and agreed to in

writing by the companies. Each company then made investments in asset specificity.

15

CSU trained its personnel to enforce the standards at each store when Hortifruti delivers
the produce and also CSU has personnel at Hortifruti’s distribution centers supervising
the standards and the orders for each store. Hortifruti has one person at each CSU store
ensuring that produce inventory does not run low, providing produce display advice,
helps in stocking the shelves and forecasting needs and replenishing orders. That is,
Hortifruti acts as a “channel captain” for CSU’s produce section. This reduces product
and time waste for both companies and improves the quality for the consumer, thus

expanding the market.

Second, Hortifruti made substantial investments to centralize product reception.
It built two distribution centers (DCs), one in Managua (the first production zone and
area of most stores) and one in the northern city of Sebaco (the other main production

zone). Hortifruti then delivers produce from the DCs to the stores.

Third, for various products, Hortifruti gradually developed since 2000 a
preferred supplier system similar to its Costa Rican operation. In the case of tomatoes,
it “listed” 43 suppliers who supply all of their tomatoes. This is a small group, but
supermarket produce retailing is just starting in Nicaragua. They still go to the

wholesale market and import from Costa Rica and Guatemala to fill any shortfalls.

There are several reasons they moved to the system of buying directly from
preferred suppliers locally. (1) It saves transport costs that had been incurred in their
importing from their preferred suppliers in Costa Rica. (2) They can work out
production calendars with the suppliers to get more even seasonal flows of product,
solving the dearth/glut cycle they faced in the wholesale market. (3) They can provide

price incentives and technical assistance as a “private good” the investment in which is

16

repaid by having contractually captive suppliers. That also then repays asset specificity
on each side. (4) It cuts out wholesaler margins thus saving costs, hence Hortifruti itself

becomes the only wholesale link and captures the profit from that.

The relation between Hortifruti and the preferred suppliers is an informal
contract (a verbal agreement and a list maintained over time); it is a contract in the
sense of Hueth et al. (1999), where a contract is defined broadly as an agreement where
there are costs (reputation, coordination, and other) incurred should one or the other
party break off transactional relations. In Table 1, this is indicated as a Specifications

Contract strategy.

The contract includes standards for product quality, and specification of
transaction attributes (such as timing and volumes of deliveries). Hortifruti only accepts
tomatoes that meet CSU’s standards; farmers have to grade their tomatoes, and deliver
the top (commercial) grade to Hortifruti and (himself/herself) sell the “seconds” to the
wholesale market or alternative markets. Hortifruti pays a price some 10% above the
wholesale market price (controlling for quality), essentially as a reward for the grading
labor done by the farmer. Note that on the one hand Hortiftuti watches the public
wholesale market prices and pays a premium to avoid that its producer sell the tomatoes
to brokers or other buyers because of temporary high prices (moral hazard), but on the
other hand, Hortifruti secures a minimal price that would help a farmer stay in business

when prices at the wholesale market are low.

To support the farmers’ capacity to meet the above contract requirements,
Hortifruti does several things. (1) Hortifruti provides some technical assistance to its

farmers - through its agronomists who visit the farmers to recommend tomato varieties

l7

(for shelf-life), irrigation systems and water sources, and crop calendar planning (for
quality and consistency and multicropping to facilitate availability throughout the year
and avoid the glut/dearth cycle of the traditional market and farmer), use of stakes and
strings (for color) chemical doses and types and other practiCes to meet their quality
standards. ). (2) Hortifruti also provides some selected credit (for new entrants), and
transport of the crop (although it provides price incentives for farmers to use their own
transport).Taken together, the technical assistance, credit, and transport “grants” form
an interesting example of “tied” credit based on delivering output to Hortifruti, and is
thus an inter-linkage of technical assistance and credit markets, the output market, and
the wholesale sector, addressing idiosyncratic market constraints facing the farmersz.
However, the method of addressing market failures (for example, lack of credit and
technical assistance) for this specific group of farmers is based essentially on donor
funding, external to the resources of the firms and farms involved. In other situations
where this occurs (outside projects) it is often the client firm with “deep pockets” that
finances the same sort of actions.3 This latter case in fact forms part of the situation in
the relation between farmers and Hortifruti, this time mediated and facilitated by NGOs,

as follows.

While Hortifruti provides the (limited) above assistance, in reality the great
majority of the assistance facilitated/arranged by Hortifruti for its preferred suppliers is
actually delivered by yet another set of actors, the non-govemmental organizations
(N GOs). The NGOs provide services of extension (technical assistance) and financial
services (grants for equipment and loans) to many of the farmers (the smaller farmers,

as shown in the next section) on Hortifruti’s supplier list. Hortifruti “pays”(in kind, not

18

in cash) the NGOs by providing information on market requirements and needs,
planning together to reduce coordination failure risk, and agreeing to buy product from
these farmers, which in turn increases the probability of the NGO’s program success,

defined as successfully implementing their “farmer to market linkage” programs.

In sum, in the upstream link, wholesaler-farmer, for the leading retailer’s supply
chain, Hortifruti has shifted from sourcing from the spot market to use of a preferred
supplier relation, to better coordinate the timing and quality of supplies from the farms in
order to reduce coordination errors related to supply shortfalls and waste from poor
quality, and to avoid the transaction costs of re-selecting (often necessary after buying
from the wholesale market) or selecting from mixed grade purchases from the farm (as

they buy only the high grade and the farmers have to do the selection labor themselves).

Second- Tier Supermarket Chain (La Colonia) tomato supply chain.
Although La Colonia has fewer stores than CSU, La Colonia’s stores tend to be bigger
and target hi gher-end consumers, and at the moment of our study both chains sourced a

similar total volume of tomatoes — but La Colonia used a different procurement system.

First, La Colonia uses a mixed system of produce procurement, partly
centralized (using their headquarters in the main store as the distribution center), and
partly decentralized (each store sourcing separately). The'chain was considering an
investment in (but at the time of our study did not yet have) a distribution center, a
costly investment. (We gathered that the chain considered there to be competitive
pressure to adopt distribution technologies of the leading chain.) The centralized system

is used for products of higher value (where quality monitoring is important) and of

19

large volumes (such as tomato) where centralized reception allows for economies of

scale in handling and distribution.

Second, La Colonia uses several wholesalers per main product, with some
wholesalers purveying several products. Table 1 shows that'the chain works with two
preferred-supplier wholesalers for tomatoes; they fit our definition of new-generation
wholesalers. The coordination mechanism between La Colonia and the two wholesalers
can be characterized as a “relation — based alliance vertical coordination strategy,” with
no equity exchange involved. La Colonia set the standards and quantities to be supplied
by each wholesaler and the amount received from each wholesaler depends on the
volume they can supply at the standard required. The two wholesalers have long (7 and
13 years) worked with La Colonia, but the relation changed markedly around 2000. The
main shift was from relatively loose application of standards (using the wholesalers
more nearly as just selected traditional wholesalers) to more strict standards (emulating
those of CSU/Hortifruti) and stricter enforcement including discounts fOr low quality

and temporary reduction of quotas as castigation.

Third, La Colonia’s two tomato wholesalers traditionally just bought from other
wholesalers in the wholesale market in Managua and sorted for different clients. But
since around 2000 they have developed a preferred-supplier system with 26 farmers
from which they get the great majority of their product (filling in shortfalls from
wholesale market). The wholesalers select the best grade tomatoes from these farmers
and then sell the seconds (at a price 20-30% lower than the firsts) to the wholesale
market. We characterize this coordination strategy as a hybrid between a specifications

contract and a relation-based alliance, in transition toward the latter.

20

Why did the wholesalers move to this new coordination arrangement? The
wholesalers explained to us that the transaction costs are smaller buying directly from
farmers and then sorting, than buying from brokers in the wholesale market and then
sorting. Importantly, in order to reduce as much as possible the transaction costs, the
wholesalers have an incentive (and we show in the data analysis that indeed they do) to
work with farmers that are somewhat larger and more capitalized than the traditional
farmers, in order to have a greater probability of getting lots of top grade fruit from any
given farmer. (They do not have NGOs working with them to help smaller farmers to
fit that profile.) Also, they note that the new arrangementreduces risk: price fluctuates
a lot in the wholesale market, leading to sudden losses, while they can have a more

stable price agreed with the farmer.

Why did the farmers accept this new coordination arrangement? The farmers
working in this system can get a good price and reduce transaction costs and risk
working with traditional brokers. This system is less demanding of the farmers because
the wholesalers buy the gamut of grades and sort, while in the Hortifruti system the
farmers themselves have to find brokers to buy their seconds.4 Moreover, the two
wholesalers help the farmers by facilitating (as a third party) the provision of input
credit to the farmers via market-based deals with the input suppliers. The wholesalers
have alliances with input firms; they have a credit line with the firms who in turn
provide inputs to the farmers; the wholesalers guarantee payment (hence the wholesale
contract is the “collateral substitute” for the input credit of the input firm to the farmer).
The arrangement is that of “tied credit” in that the farmers have then to deliver their

tomatoes only to the wholesalers. This is an interesting example of inter-linked credit-

21

input-wholesale markets to overcome idiosyncratic market constraints for the farmers

(who lack collateral to get the inputs) such as credit and technical assistance.

Given that La Colonia’s and CSU-Hortifruti’s systems share some similarities,
why doesn't La Colonia use Hortifruti, or CSU use La Colonia's two wholesalers? La
Colonia likely needs to be differentiated from CSU, and Hortifruti is connected to CSU
in the holding company and in other countries. Note in sum both chains use of preferred
supplier lists (long term relations), as well as linking the output and credit market (in
various ways, direct and indirect) and facilitating farmers with some credit and
technical support. The main differences are that farmers have to sort for Hortifruti, and
get NGO technical assistance in the Hortifruti chain, and get credit from private sources
with the La Colonia system, and from NGO sources as well as from the wholesaler in

the Hortifruti system.

The Traditional Retailers ’ Tomato Supply Chain. We take the supply chain to
the main public wholesale market in Managua as typical and also by far the point of
greatest traded volume of tomatoes in the country.5 There are 10 tomato wholesalers

operating in this market.

The traditional wholesalers by definition sell only to traditional retailers. Between
traditional retailers and the wholesalers, the relations are a typical spot market - price
discovery, then decide whether to make transaction. We asked the wholesalers why they
do not sell to the supermarket chains. Several answers were common. (1) They do not
want to spend the time sorting tomatoes or monitoring farmers who would sort the

tomatoes. (2) They do not want product rejections (discounts) from the chains and the

22

transaction costs these imply. (3) They also do not want to experience the delay in
payment (receiving the product and then paying the supplier say in several weeks) which
is the typical practice of supermarket chains. (4) Many felt the supermarket chains are
still too small and buy relatively small volumes — thus implying high average fixed costs

dealing with them.

Wholesalers buy either from rural brokers (who source from farmers in rural
areas) or farmers who come to the market, again only under spot relations. Whether
directly to wholesalers or to a rural broker, farmers in the traditional channel merely
sell ungraded (and with no standards imposed) their tomatoes, for an average price (the
same traditional system as one sees in many developing countries). The wholesalers sell
the tomatoes to retailers either ungraded or in some cases graded by quality (size and

degree of damage).

Finally, in the traditional retailer supply chain analysis note that traditional
wholesalers and retailers appear not to face costly coordination errors from spot market
relationships given that their customers probably care more about price than quality in
the purchasing decision, and both CSU and La Colonia supermarket chains are
strategically committed to differentiating themselves in the market place thus

increasing complementarity and asset specificity in their respective supply chains.

3. PRODUCERS IN THE THREE MARKET CHANNELS
The above section showed that the interface between the buyer (the

retailer/wholesaler combination) and the farmer is different in each of the three supply

chains, and thus the three market channels facing the farmer. The grades and standards,

23

as well as the transaction attribute requirements (communicated through and enforced by
coordination mechanisms) of the two supermarket chains taken together (although the
two chains have substantially different market channels leading to them) are more
demanding than the traditional channel, and all else equal, one would expect farmers with
greater capacity selling to them (size and/or the various forms of capital such as human,
financial, physical, and organizational). That hypothesis is of course balanced against the
point made above that the supermarket channels also provide various forms of assistance
and support that taken together are more than can normally be accessed by traditional
farmers. The hypothesized effects of the vector of capacity variables are ambiguous in
that larger/more-capitalized farmers might be drawn to the supermarket channel (and
welcomed to it) for its greater rewards coupled with greater demands, but the smaller
farmers might also be drawn because of the channel’s greater provision of support such
as technical assistance and credit. Such ambiguity suggests the need for empirical work to
test the hypothesis that greater farmer capacity (land or capital or both)>determines access,
and once the channel is accessed, technological behavior. The answer to this issue is
crucial for development debates in the region, as policymakers want to know whether the
emerging supermarket channels are inclusive of small and poor farmers, or sidestep them,
and if necessary what type of upgrading efforts may be needed to help small producers’

access supermarket channels.

In turn, one would expect that the technology, output supply, and input demand
choices of farmers will differ over channels. A hypothesis is that in general the modern
channel farmers will be more capital-using in order to meet quality and consistency

requirements, based on a similar hypothesis presented by Key and Runsten (1999) for the

24

processing sector in Latin America. It is of course possible that smaller farmers can
substitute “sweat equity” (labor) for capital and thus participate in the modern channel,
for example picking weeds instead of using herbicides, watering by hand instead of using
irrigation, and so on. The possibility of substitutability in factors (labor by capital) is
central to the policy debate because if the smaller farmers can participate in modern
channels, even with low capital but strong arms, the effects of the transformation of food

markets may yet be inclusive rather than exclusive.

This section tests the above hypotheses, essentially asking how, among the
channels, the participating farmers differ - in the farm characteristics reflecting mainly
capacity variables that determine their channel choice, in the technology and input
demand and output supply choices given their channel choice, and the net returns to their

choices.

The section proceeds as follows. First we discuss the data, survey, and sample.
Second, we descriptively lay out findings concerning producer characteristics in the
various channels, and the net payoffs to their choices. Third, we use the endogenously
stratified sample to test for differences, in turn, in technology and in output supply/input

demand behavior.

(a) Data and sample

From our qualitative interviews we learned that the supplier lists of the
supermarket chains contained few farmers compared to the many traditional channel
farmers. We thus did a reasoned sample based on the supermarket suppliers’ population
and the traditional channel population. We obtained the lists of farmers selling to the

wholesalers of the two supermarket chains, and selected that whole universe; this

25

included all 43 producers selling tomatoes to the CSU channel, and all 26 growers selling
to the La Colonia channel, for a total of 69 “supermarket channel farmers.” Further, we
sampled at random 76 growers chosen from a list of 300 producers supplied by a random
sample of two of the 10 wholesalers in the Managua wholesale market. We limited the
traditional-channel population to 300 so we could have a sampling error of 10% with a

probability of 90%.

The data are from a farm survey conducted J une-July 2004. Each interview lasted
about one-two hours and posed quantitative and categorical questions concerning
household and farm characteristics and production, organizational, and transactions
behavior, as well as qualitative questions concerning their perceptions of the alternative

market channels.

(b) Farm Characteristics, Technology Use and Profits

In this section, we first present salient characteristics of the producers
selling to the three supply chains (recall that we use that term now interchangeably with
market channels as the latter is the supply chain viewed from the farmer perspective).
Besides stratifying the sample by market channel, we also divide it into farmers working
with NGOs and those not. This further stratification, rarely seen in the literature, allows
us to see what sorts of farmers, in each channel, participate in the market free of donor-
supplied assistance. That is a crucial issue in the policy debate because the debate will
take different directions over the next few years whether one finds that the smallest

farmers can participate in the modern channels only if (heavily) assisted.

26

The subsection draws on three tables: Table 2 shows the characteristics of the
farms; Table 3 shows breakdowns of production costs; Table 4 summarizes technology

and profit indicators. Several salient points emerge.

First, the great majority of growers for CSU-Hortifruti are assisted by NGOs -
although half of the tomato volume bought by Hortifruti is from non-NGO-assisted
growers. This shows a key point: in an “unassisted externally” situation, the leading
supermarket chain relies on farmers who can sell a larger volume each of tomatoes, and
works with smaller growers with NGO support. We noted in section 2 that the NGOs act
as a technical assistance arm for the leading chain - but not just any “generic” technical
assistance (as might be got, if it is even available, from extension), but technical
assistance geared to helping the smaller farmers meet the specific norms of Hortifruti.
The NGOs helped growers to form cooperatives (or assisted in improving organization
for those already in groups), seen in the high degree of association participation of the
assisted-growers as compared to the others. This was necessary given a'fact that leaps out
from Table 2: the average tomato area of the assisted is four times smaller (and produces
13 times less volume of tomato) than the average non-assisted grower. The NGOs thus
work with producers that are somewhat poorer with smaller outputs, via assisting their
coops with substantial credit and technical assistance, to raise incomes and help them

diversify away from basic grains.

By contrast, very few growers for La Colonia are assisted by NGOs, and where
they are, the assistance targets production activities and is not delivered in an “alliance
relationship” (which is the term for the NGO relationship with the coops assisting

Hortifruti) to help farmers enter the supermarket-market per se. Yet the tomato area of

27

the average farmer among the few La Colonia NGO-assisted farmers is four times that of
the assisted farmers selling to CSU. That is, the assisted and unassisted farmers selling to
La Colonia are already large enough (in tomato production) to qualify for the preferred
supplier lists of the wholesalers for La Colonia. Recall from "Section 2 that the
wholesalers for La Colonia arrange credit from private input suppliers (who also give
technical assistance at least for their chemicals) to the farmers.

Second, the overall farm size of producers selling to both supermarket chains is
not statistically significantly different from those selling to the traditional market.
Indirectly this suggests that farm size is not an entry requirement. In any case, farmers in
Nicaragua selling vegetables, including tomatoes, to the local market, whether modem or
traditional, tend to be small, so supermarket chains and wholesalers are basically limited
to working with the spectrum of what can loosely be classed as small farmers, who in fact
vary sharply in terms of capital, crop composition (hencetomato area), and even farm
size substrata within the small farm base. This kind of differentiation over small farmers
has for several decades been noted in the Latin American literature (see Schejtrnan 1996).

However, the tomato production scale (tomato area) of producers selling to La
Colonia is much bigger than farmers in the traditional market; by contrast, the farmers in
the Hortifruti/CSU channel, in particular the small, assisted growers, have much smaller
volume per farm, compared with the traditional channel farmers. At first glance, this
difference in implied-required-scale of farmers between the two supermarket chains is
striking. But a second look removes the surprise: the assisted growers are grouped (with
NGO help and guidance) into three cooperatives each of which acts as a medium supplier;

the NGOs even work with the coops and Hortifruti on planning of output (cropping

28

calendar). The picture that emerges is then that each of the supermarket chains ends up
sourcing from the larger-stratum of small farmers or coops delivering a volume as large
as a medium farmer would deliver. Seen that way, the striking point is rather that the
supermarket chains rely on suppliers (individually or collectively) who sell them
substantially larger volumes per supplier than could be got from the traditional farmers.
Third, a crucial point is that the non-assisted growers for La Colonia are in many
ways similar to the non-assisted growers for CSU as compared to NGO-assisted growers
selling to both chains. The non-assisted growers of both chains also tend to be more
specialized, as they produce much higher volumes on a much higher share of an average
farm, compared to the NGO-assisted growers. Interestingly, the Non-NGO assisted
growers also tend to be less experienced in tomatoes but more educated as compared to
the NGO-assisted growers. They tend to have tractors and vehicles and other indices of
capitalization, relative to the NGO-assisted growers. This is a key result as it indirectly
provides the scenario of what types of producers are able to sell to the supermarket chains
with and without NGO assistance. It reflects that without the support/subsidy provided by
the NGOs, supermarkets tend to source from growers who have more capital, are less
experienced but more educated, and are more specialized in commercial horticulture.
Fourth, as shown in Table 3, growers that are in the preferred supplier list of
supermarkets have 30% higher production costs than those selling to traditional channels
only. After breaking down the cost expenditures of non-labor variable inputs into prices
and quantities, we see that the CSU channel growers have higher fertilizer use (quantities)
but lower use (from 1/3 to 2/3 less) of quantities of other toxic chemicals (insecticides,

fungicides, herbicides) than the other supermarket channel - but both chains’ growers

\

29

have higher than the traditional channel, thus confirming Thrupp et al. (1995) argument
about modern retail (in her work it is developed country retailers importing fruit from
Central America) driving higher pesticide use in countries with a limited regulatory and
almost non-existent enforcement capacity, such as Nicaragua. The fact that the cash
outlays are higher is explained by Hortifruti’s requiring the farmers to use certain kinds
of chemicals, and those kinds have higher prices, reflected in the higher outlay than the
other channels. Note that producers selling to Hortifruti, both assisted and non-assisted
producers, received technical assistance concerning chemical use to meet CSU’s quality
standards —- and both groups tend to use similar quantities of chemicals per hectare,
except for fertilizers and insecticides - but that their use is substantially different from the

quantities used by the producers selling to La Colonia and traditional market channels.

Fifth, however, Table 4 shows that within the CSU/Hortifruti channel, the
unassisted farmers have 30% lower production costs than the NGO-assisted producers,
who have the highest fertilizer, seed, irrigation energy, and labor costs cf any producer
group. Given that we control for channel in making this comparison, it appears due to the
NGOs’ technological package recommendations, geared to minimizing risk of not

meeting standards.

Sixth, producers selling to CSU use substantially more labor days per hectare
because of the additional requirements as compared to producers selling to La Colonia
and to the traditional markets. This was expected and can be safely attributed to
Hortifruti’s agronomists recommendations to improve product quality such as more

intense pest control techniques (reduce insect damage) and labor allocated to use stakes

30

and strings (important for color), labor requirements of timely harvest techniques (when

tomatoes meet the standards) and labor used in grading the tomatoes.

Note that producers selling to CSU use a more labor-intensive technology as
compared to producers selling to La Colonia and to traditional markets. By contrast,
producers selling to La Colonia use a capital-intensive technology, relative to those
selling to CSU and to traditional markets. This is in part because of the additional labor
tasks requested by the buyer as noted above, and because producers selling to CSU-
Hortifruti face cheaper labor as shown in Table 3. This may be due to CSU producers’
closer location to urban centers where population densities are higher and labor may be
more available, compared to producers selling to La Colonia and traditional markets.
Also, La Colonia producers have more members of the household with off-farm jobs
which may compete for labor time and at the same time provide own-liquidity for capital

investments on farm.

Seventh, as shown in Table 4, profit per hectare is substantially higher for both
supermarket channels compared to the traditional channel; for the sample-weighted
average, supermarket-channel farmers earn 65% more profit per hectare. A small
advantage in prices paid by supermarkets, and a large advantage in average yields,
compensates for the higher outlays for inputs. Note that La Colonia suppliers earn 56%
more profit per ha than those supplying CSU. However, comparing non-NGO-assisted
growers in each, one finds only a 12% difference. What drives the difference between
these results? On the one side, experience seems to play an important role in explaining
the difference between the two chains Non-NGO-assisted growers as La Colonia

producers have on average 13 years of experience while the CSU have only slightly

31

above two years. Non-NGO-assisted farmers, desired by the chains for their prowess and
capital, have much higher yields than the NGO-assisted farmers. However, even the

NGO-assisted farmers earn 14% higher profits than the traditional growers, but doing so
depends on the general assistance and credit they receive from NGOs plus their ability to

pay 36% higher input costs up front, a significant capital entry barrier.

(c) The Determinants of Producer Channel Choice

This subsection and the next treat the determinants of grower market-channel
choice, and the correlates of those choices (technology, and output supply and input
demand). We use a two-stage approach with channel choice first (in this subsection), and
then (in the next subsection) model the correlate behaviors with endogenous stratification
of the sample into the channel strata, controlling for the conditional probability of

inclusion in a given channel.

The producer’s market channel choice can be conceptualized using a random
utility model (RUM). The RUM is particularly appropriate for modeling discrete choice
decisions such as between market channels. It is essentially an indirect utility function
where an individual (with specific characteristics) associates an average utility level with

each alternative (such as a market channel) in a choice set.

In our context, a producer for the traditional channel does not sell to the
supermarket channels; a producer for one supermarket chain does not (in practice) sell to
the other but does generally sell his/her seconds to the traditional channel as well, and

here is categorized as a grower selling in one of the supermarket channels. Hence there

32

are three mutually exclusive channels into which producers are mapped, the

C SU/Hortifruti channel, the La Colonia channel, and the traditional market channel.

The producer’s utility from participating in market channel j is represented by U3.

The producer makes a marginal benefit-marginal cost calculation based on the utility

achieved by selling to a market channel or to another.

1 U..= .X..+£.. V'eN
() U fl] U U J

We can not observe directly the utilities (or the difference between benefit and
cost) but the choice made by the producer reveals which one provides the greater utility

(Greene, 2000).
A producer selects market channel j=k if

(2) Uik > Uijvj' ¢ k

where Uik denotes a random utility associated with the market channel j=k, and ,6]. X if is

an index function denoting the producer’ average utility associated with this alternative.

The second term 61.]. denotes a random error which is specific to a producer’s utility

preference (McFadden, 1976).
Now, in our implementation model, market channel choice is modeled with a two

equation system (j=1, 2) as:

33

3 Y..= .X..+ ..
() U '6] I! 81]

Where in the first equation Yn=l if producer chooses the CSU-Hortifruti market

channel (Leading supermarket chain), 0 otherwise (which is'equivalent to Yil = 1 if

U.
r

1 > U ij Vj at 1). In the second equation, Y,2=l if the producer chooses the La Colonia

market channel, 0 otherwise. The Managua public wholesale market channel is the base

category for both equations. ,6]. is a vector of channel—specific parameters. 8ij is the

error term assumed to have a logistic distribution with mean 0 and variance 1.

X,,- is a vector of producer characteristics that together reflect the incentive and
capacity variables influencing the producer’s indirect utility, hence his/her market

channel decision, and includes the following variables:

(a) F ARMSIZE: the overall farm size in hectares; this is measured as the sum of
land owned plus usufruct, rented-in, and sharecropped-in land, minus the sum
of rented-out and sharecropped-out land in the cropping season of 2004. We
expect a positive effect of this variable, as it is linked to marketable volume

considered desirable (by the buyer) as it reduces transaction costs.

(b) LIVESTOCK is the quantity of livestock owned by the producer and is used
as a proxy for household non-land wealth, and hypothesized to positively

affect choice of the supermarket channel.

(c) TRUCK is a dummy variable equal to one if the producer has a truck and 0

otherwise. This variable increases the capacity of the producer to deliver

34

tomatoes to the supermarket distribution centers according to a specific

delivery schedule and possibly from distant regions.

(d) ROAD is the distance to a paved road measured in kilometers, expected (as a

measure of transaction costs facing the producer) to negatively affect choice

of the supermarket channel.

(e) ELECTRICITY is a dummy variable equal to one if the producer has access

(0

to electricity, 0 otherwise.

HOUSEHOLDSIZE is the number of members in the household during the
last six months. DEPENDENCYRATIO is the dependency ratio is the number
of members in the household with less than 15 years or older than 65 years
over total household size. FEMALE is equal to one if the producer is a female,
0 otherwise. AGE is the number of years of the producer. These variables
reflect characteristics and the availability of the labor in the household. All
else equal, we expect more adult labor to favor selection of the supermarket
channel as the various extra steps (relative to the traditional channel) in
production and post-harvest require more labor. Gender and age are included

as proxies of the producer’s indirect utility preferences.

(g) EDUCATION and EXPERIENCE are the years of schooling and the years in

tomato production of the producer. We expect them to favor entry into the
supermarket channels, which are more demanding in coordination of field and

post-harvest activities and chemical use.

35

We use a multinomial logit regression to estimate the equations because of its
simplicity and because it is a model that fits multiple discrete dependent choice variables.
We estimated it using the weights suggested by Manski and Lerman (1977) and Cosslett

(1981) to correct for sampling bias.6

The market channel participation results are presented in Table 5. Several

significant findings emerge.

First, importantly, overall farm size was not a significant determinant of selling to
the supermarket chains. This contradicts our expectation that farm size would be a key
determinant. But looking back at Table 2, one sees that more than farm size, the degree of
specialization in tomatoes is important. This coincides with qualitative discussions with
the wholesalers who want farmers committed to tomato farming (and thus also with asset
specificity which makes them more committed to the relationship with the buyers. . .).
Moreover, as we noted above, smallness of the farm’s output is balanced by the

associations among tiny farms.

Second, choosing the CSU/Hortifruti channel is positively correlated with more
education and negatively correlated with experience. The education result is obvious,
but the experience one may seem odd until one notes that many of the assisted farmers
are newly in tomato production, having been helped to switch to tomatoes from less
profitable crops. Moreover, older producers (controlling for education and experience)
tend to sell to the traditional channel. It is possible that youthful energy and

entrepreneurship is needed to brave the step to the modern channels!

Third, interestingly, selling to supermarket channels is positively correlated with

distance to an asphalted road, which indicates that the chains may not just work with

36

producers with better road access. In other words, those producers located in the
hinterland may still be part of the chains preferred supplier lists. Presumably, as one
would expect the more the tomatoes are bounced over unpaved roads, the higher the
loss rate. However from our qualitative interviews we learned that logistic
improvements were introduced to minimize losses, such as a maximum weight per
plastic crate and other measures that reduced the contact between plastic crates. On the
other hand, the difference in average distance to a paved road for those farmers selling

to supermarkets versus the traditional market does not appear to be considerable.

(d) Comparison over Channel Strata of Technology and Efficiency
Here we estimate the production functions for the different channel strata.
Subsequently we derive the technical efficiency and allocative efficiency results.
Conditional on market channel choice, the producer’s production function is represented
as:
h a j h

(4) Y..=A..HX.

I] U Uh VjeN

where Yij represents the ith farm’s total tomato production in kilograms in 2004, jth

market channel. In the technological relation Ajj represents the technological efficiency

level. ajh is a vector of positive parameters. X,,- is a vector of h inputs. As endogenous

stratification is used, and this is the second stage of the switching regression the first
stage of which is the channel choice, we include in the estimation model the fann-
specific observation for the producer’s conditional probability of being in that channel

(represented by the inverse mills ratio (IMR) calculated from the first stage). We use a

37

logarithmic transformation of the Cobb Douglas function; there was no additional
explanatory power from use of the quadratic or translog functions, as judged by a
standard F-test. 7 We estimate the production functions using its standard log-linear

transformation as follows:

+...CZ .

(5) lnYij =lnAU+ajllnXi. jh

11 lnXijh +ajlij +17”.

VjeN&‘v’heX

Where the vector Xi,- includes the following variables, all totals for 2004:

(a) LABOR is total labor days used to produce tomatoes; it includes family and

hired labor.
(b) FERTILIZERS is the total cost of organic fertilizers.

(c) F OLIARS, INSECTICIDES, F UNGICIDES, HERBICIDES, SEEDS

represent the total cost of these items applied.8
((1) LAND is total tomato area measured in hectares.
(e) IRRIG is the percentage of tomato land that is irrigated.

(f) AGMACH is a dummy variable equal to one for use of tractor and 0 for

animal traction.

(g) REGl and REG2 are dummy variables for regions, which aim to capture

differences in soil and weather conditions between the regions.
(h) xii]. represents the inverse mills ratio (IMR) calculated from the channel

choice equation, used to correct for selection bias. Following Heckman (1979)

38

we use an extension to multiple outcomes further discussed in Vella (1998),

Dubin and McFadden (1984) and Bourguignon et al. (2004).

(i) a]. is the vector estimated parameters

(j) 771.]. is the error term9.

The regression results are shown in Table 6. We applied the Chow Test to test the
null hypothesis of equality of production function parameters of the three supply chains

as a = a}. ¢ ij and the alternative hypothesis of at least one parameter

1' = k
statistically different. At 99% significance level the null hypothesis was rejected in all
cases"). Hence, the results support that the technology used by the producers in the three

supply chains differs. 1'

We use the parameters to examine efficiency. The producers selling to the
supermarket chains are substantially more technically efficient than the traditional
channel growers. Lau and Youtopoulos (1971) note that technical efficiency is reflected
in the intercept term of the Cobb Douglas production function. A comparison of the
intercepts of the production functions as measures of technical efficiency indicate that
producers selling to La Colonia have the highest technical efficiency, followed by
producers selling to CSU, with the traditional growers the lowest. In terms of partial
measures, the marginal physical productivity (MPP) of land is three and six times higher
for producers selling to the CSU and La Colonia supermarket chains compared to the
producers selling to the traditional markets. Moreover, the MPP of organic fertilizers and

chemicals of producers selling to the supermarkets is higher than that of the producers

39

selling to the traditional channel. Producers selling to the traditional channel show a
higher MPP of seeds as compared to supermarket chains, which suggests that they might
have difficulties in acceding to either good quality or enough quantities of seeds. The
MPP of labor of producers selling to La Colonia supermarket chain is highest and it is an
indication of less labor available (as noted above). The MPP of labor of producers selling
to the CSU supermarket chain and traditional markets is somewhat similar. These results
taken together are probably due to the greater education, specialization, and assistance

received by the supermarket channel producers.

Using the physical productivity estimates plus product prices and factor prices,
we examine allocative efficiency. In neoclassical theory, allocative efficiency occurs at
the point at which the marginal value product (MVP) of a factor equals the factor’s
price. If the MVP exceeds the factor price, more of the factor could be profitably used
and this reflects a constraint to access of the factor, and the opposite, an overuse of the
factor. For example, Carter and Wiebe (1990) show that smaller farmers in Kenya have
a constraint in access to capital (such as fertilizer) but an overuse of labor (indicating a

constraint to selling labor in other markets).

Table 7 compares the MVPs with the factor prices. Here, as in the next section,
in order to avoid endogeneity, we use average prices (observations for each of the seven

municipalities) for factors and output. The main results are as follows.

First, producers selling to CSU-Hortifruti are close to being allocation-efficient
in chemical foliars, fungicides, and seeds. The MVP exceeds factor prices for labor,
fertilizers, and herbicides, but not by much. However, the MVP of insecticides is below

their price, indicating overuse. This practice is common in agroindustry outgrower

40

schemes, as the buyer is sensitive, as in the present case, to insect damage, so producers
overuse pesticide to reduce risk of rejection. The MVP of land is well above the land
price; this land constraint may well be due to lack of irrigation area and the low quality
of the non-cropped land in their holding (mainly hillsides and rocky areas), as well as

land market constraints. '2

The producers selling to La Colonia and to traditional market channels exhibit
excess of MVP of labor over wage, indicating a labor constraint, corroborating points
made above. Both sets of these growers also over-use chemical inputs (the MPVs of
these factors are well below their prices). They may be compensating for the labor
shortages, as well as (over) managing risk from pest infestations, and lack of

knowledge from limited technical assistance.

(e) Output Supply and Input demand functions

Here we estimate the output supply and input demand functions, again
endogenously stratifying for the three market channels and controlling for selection bias
by including the IMR in the equations. The equations are derived from the profit function
using Hotellings Lemma (without needing the assumption of profit maximization

behavior, see Sadoulet and de Janvry, 1995) and obtain the following three equations:

(6) =6.+6.P +5.W +6.K..+6./1..+n
J m J m J 1J J 1J

in(marketings) j ij

=a.+a.P +a.W +a.K..+a.}...+77..
J m J 1J J 1J 1J

(7) X ji (labor) _] j m

(8) in(non—Iabor):’6j+fljpm+flij+fljKij+flinj+nij

41

Where: Yijmmkeungs) represent the producer’s output supply (no home-consumption
is assumed) during the 2004 cropping season, Xijaabor) represents the producer’s demand
for labor measured in total labor days per farm per year and includes family and hired
labor days. Xij(non-Iabor) represents the producer’s demand for‘non-labor variable inputs.l3 P
is the average price per municipality of tomatoes measured in US$ per kilogram. W is the
vector of input prices (per municipality) and it includes the wage rate (U 88 per day), and
the non-labor variable inputs price per municipality (U S$/Kg).K is the vector of fixed

factors and it includes:
a) LAND is total tomato area measured in hectares.
b) IRRIG is the percentage of tomato land that is irrigated.

c) AGMACH is a dummy variable equal to one for use of tractor and 0 for

animal traction.

d) REG] and REG2 are dummy variables for regions, which aim to capture

differences in soil and weather conditions between the regions.

e) xii]. represents the inverse mills ratio (IMR) calculated from the first stage

(market channel choice model).

1) 6]. ,orj , ’Bj are the estimation parameters and 171.]. is the error term.

We estimate the system for each market channel independently using a Zellner’s
seemingly unrelated regression (SUR) model to exploit potential correlation across the

errors in the three equations.

42

The marketing function and labor and non-labor input demand function results
are presented in Table 8. The regression results combined with average sample variable
values are used to calculate the elasticities and absolute marginal responses of price and
non-price factors of producers selling to the three market channels. Several salient

points emerge.

First, the output supply price elasticities and intermediate input price elasticities
(labor and non-labor variable inputs) are higher for producers selling to supermarket
chains than those of producers selling to the traditional channel. For example, an
increase of 10% in the price of tomatoes will increase the supply of tomatoes of CSU
producers by 777Kg, which is 95% higher than the expected increase in supply of
traditional producers. La Colonia producers will increase two times and a half that of
CSU producers. While this measures a cross-section response, it is typical in the
literature to interpret this as what would be the medium-long term response to price
increases, and this suggests that the modern channel producers are more sensitive and
responsive to price changes than the traditional channel farmers, perhaps because of

their commercial orientation and greater access to assistance and credit.

Second, the own-price elasticity of labor demand is substantially higher for the
traditional market than for the supermarket channel growers; moreover, the own-price
elasticity of capital (non-labor inputs) is much higher for the supermarket channel
growers (with the expected differences between the chains, with a lower elasticity
among the CSU growers due to the presence of the smaller NGO-assisted farms). If one
thinks of that elasticity as correlated with the slope of the isoquant (labor-capital) where

the factor price ratio is tangent, this corroborates the point above that the traditional

43

farmers’ technology is more labor-intensive, while the supermarket-channel growers

(on average) is more capital-intensive.

Third, producers’ non-price elasticities are somewhat similar for producers
selling to the three market channels, but the elasticities are relatively low. By contrast,
the elasticity of output supply with respect to land is higher for supermarket-channel

producers, expected from our findings above of higher yields of that group.

44

4. CONCLUSIONS AND IMPLICATIONS

At a very early stage of development of the supermarket sector in Nicaragua,
results indicate that the main two supermarket chains have already introduced vertical
coordination strategies going beyond the spot market relations still used in the
traditional market. The supermarket chains have sought to win middle-class consumers
with higher quality while extending their sales to the lower middle and upper-lower
income segments with lower prices for foods, and recently have started on this path
with fresh produce. The chains shifted to greater control of quality and lower
transaction costs, hence overall costs, with the very recently new coordination strategies,
involving new-generation wholesalers rather than traditional brokers, de facto contracts
with farmers on preferred supplier lists, and arranging (through their own agronomists,
or NGOs, or input suppliers) technical assistance and credit that are apparently difficult

for those merely in traditional markets to access.

The regression and descriptive statistic results show that while farmers in the
“small” category can access the supermarket channels, the farmers who do so
unassisted by NGOs tend to be the larger and more capitalized tier (in education,
experience, and physical capital) of small farmers, and those assisted by NGOs, while
less capitalized and smaller, are helped by NGOs with technical assistance and
organizational assistance to form groups that overcome their asset poverty to meet the
modern channel’s requirements. The results also show that once accessed, the modern
channel allows substantially higher profits (from slightly higher prices and substantially
greater efficiency, using more capital-intensive technology) to farmers compared to the

traditional market.

45

Several implications are pertinent. It is clear that for farmers to enter the modern
channels, they need a spectrum of forms of capital that most traditional farmers, even
commercial ones, do not have in abundance: experience, education, organization,
irrigation, access to technical assistance and finance, even sOme greenhouses. That of
course means that, unassisted, most farmers cannot access these market channels. While
the “elite” of small farmers enter unassisted, the smallest (non elite in land and capital)
enter the modern supply channel with help from NGOs -— NGOs who employ agronomists
and business management specialists usually well beyond the reach, too expensive for
access, of the traditional government agencies charged with rural development. It is
instructive to note that a representative project of the NGOs spends, by our calculations,
about 8 times more per farmer in these programs, than do the agricultural agencies of the
Nicaraguan government. It is safe to say that the Nicaraguan government either cannot or
will not be able to afford mass-scale upgrading of farmers to enter modern channels. At
present this is not a problem, as the supermarket-market is still small. But the growth of
the latter means that some day, perhaps in the next decade, supermarkets will control half
or more of the produce market, as they do already in Brazil. The development program
and policy issue of how to upgrade large numbers of farmers to prepare for the market

transformation thus emerges as a critical issue for debate today, to prepare for tomorrow.

46

.??aco mo £85092 =2. 05 v38? 28 coca-Eton 83 28:0 >393. 2: 2.00an 32333 BEES-am 2: no £80833 2: mo
bzmaao mo .0:ng .8 68 39.0 < 6x; we Ewe-«Em?- 3... .Qam “a 53$ch s... ...\oo_ 8 Eng-:35 ._ .momoficoaa 5 Echo vac—5m 38:2

 

 

 

47

:3 Se :3 3o 88 5. 285 E BE
:3 8.8 88 88 Se 3o 83.. 2:8
:3 :3 88 Se 88 So Ba SE .355
833.38 8882:.
§ 42 2 2232.030
ado a; as $3 $3 :8 2
A38: A338 Go: 8 $23: $.58 88.28
8.? 3%. 3.52. 9..:~ an: 8.2a 2328
$28 :38 638 A888 538 :38
.38. :8:- mndN- 42- :...fi.% :35- 8.5:.
$.88 E88 838 85: £38 6:8
.23. :_.-.$ 2a- 83 6.8- :83 32:5
E8 828 E8 3.8 58 A38
8.... c3 .33. 38 88- as- $8.8 8E
58 2:8 A28 398 58 A38
8.? :85 :5 :28 88 .23 :85 E 8E
88.8 88.8 298 §8 9.8 5.8
88 88 3o :3 :E88- .888- 8E 2.3
a». :8 838 $38 338 6&8 $38
8.4m 3.3 3.3. 8: _- w _ . _ m- :83: SE 380
«3396 8:9...» 3:395 3:925 8:993.
0.3.5 28.5 9.3.5 0.3.5 3.39.... 23.5
San—oh 2:: 25:0...— o_a2 0.3.5 0.2.5..— o_a2 mouafizmm c8252!—
.a:o_=ta._.—. commune...— «oxuafihonaw

 

Naomugauaouv 653:0 Coo—SE .3 829%; 20260.5 “2 033-

 

 

:38 $38 828 828 5.28 5.8.8
NE- 3. _ m- 8.2- 2.? :2 2m $.9- 8Es 8:85
an: :3. 8 A38 3. 8 8:8 A88
:3. 85 88 $3- «.3 N2 :58 82 8:213 9 852.5
538 case as. _ 8 5.5 3:8 838
:38: :28: 8.: 8.3 :NSS .33: 5:88.:an
:58 2:8 A88 2:8 g8 2 _.8
.23 523 85 32.9 3.3.... a; 82.. 2:6
2 _.8 :58 A58 938 588 A88
8.? .85 :88- 2:38 .858 :88 Em .EE .335
A: .88 838 33.8 338 5.38 22.38
San 2:3th EN- 8.3 :38. :asam 2am 55285
5.8 3.8 5.8 858 $8 $28
as- 2.33.2 E.» 43 new. 8d Ba 2238:
$98 38.8 358 G_. 8 5.8 898
8.? ”we 53 _~._ 8._ a: BEBE 2:8 5 8:233
82: $38 83: :3 8 808 638
.36..- MEN- 3:- :.R- :32- N281 ..2383 538E
5.3 838 838 3.88 828 82m:
8.? 82$. 3. :- NEE- $3 8.2:- soéea 225“.
E8 5.: 5.8 so: 628 A28
3o- em. _- as .5. _- .334- .3?- EEE co ow...
8:893. SEE—5 8:23. 8:.53 8:25.
23:0 23:0 23.6 33.5 82.5.5 9.2.5
22:0...— ofi—z 0:25.: 9.22 0.3.5 0.2.3.: 232 moan—=33— 3.2.3.3.—
..EeE—zzh cow-88...— CoxcaEnoa=m

 

Eocene 6x88 .3 mom—mafia .mcoozcem ”2 033-

48

.bowufio .255 .5on 2.. Eco wow: 28 22222.0 $.53... 8.. baa nausea. .255
3me 03 A3 3.8% .0585 8: 25 can =0.on 2.. a. meme?“ “coco. an... 8 80.... 3.3 03 A0 .2520 .832: b. 8.83.8 .aEEs 033%».me 8:
ca 03 3.2.9:? 3.. .om.< .3338 5.8.5 9 858 :8 8.x .35 50.3—52.3. 8 30on 9.2.8 £02605 .md. 22:...“ no 890 2.20%? :0... $2.55 8.3:.
0.92:5 8 :8 2832.. 8:82. $8. 9 gm 8: ow 89359.2. 2.... AN. .32. c. 3532: 28 0.38 .59 an 383% .0555 5.83 2.. o. 8 .0585
.8885 2: 8 .o .2893 8.36.2.3 2.. 2 20m banana mo 25% o... m. 28 v.88 .3 .35553 (.o 085 2.... .8892. watofiwzufi 8.. 28 35:2? 2.2.
3.5.0:. a. 98 E83 bEaE 0.5.0:. 8.3 0.2: 05 8?. .o:.m> ozocow .o o:.m> 3.82.852 B. 8: EB 33> .32. 3. 60:2 8 8883 9.5235? .8 28
.2652? 3.5.0:. 08 auto: 98 $58 35.2: 83.2.. bemoan”. 0.25.. 2.... .8892. wEBEwssm .8 Batman 238 .82 .5352 3.3 80:... 2. 68:7.

 

 

an 3 S. an o . m 3... $8.52 8.83 E2“.
a,“ m. M mm a o. m. 3% 8.3.3.585 EB”.
a.“ a a e a 2 an 3 3... 52.8... .0523 so;
who»: .3 3.58
cm 3 co. m m an on A3 Aunts.— ..c $8 988... ~9:. 22.. Stat
0.3m ”3.“ 3:. b _ v a EM 8.. m _ v av: Ems)»
3. R 2 z 3 mm X 95.58 ow<
2% 9.08... .c 2.993 E... ow<
.0 8...;
:0 3.2 5525‘ .8339..— xhafiuonam .8525. 3389..— 5.395
Hm 3m SE 880 «3.23.2

 

Acouascucouv .2820 E 0.68 no 83589 28 3.3 Eco—60$ H: 2%...

49

 

Rm :m N; em 5. am €353: 3:2 2.25".
mm m «.3 an». no». o _ m gm €8,596 3.3 2&2
A3 $3.523 13.— ..2— 3th
a a o 2: a o 2: 3“ Q: 3888;
- - o o 2: o o 2: 3.: 2&3 E3825
5 _ _ on 2 h .2 3.: £2.23
a : mm 3 2 cm 2 ax: 35:55:35
ad ad 2: oo— _ N 2: w m $3 2283‘
N o o a _ _ E ax; £855 o>= :83 :2: 30:3
E
35335.:— uo <5 .9535 .322: .3 mafia...—
AaczoEQov 5:269:—
eé b... m..— vd m6 v... 588:. E can. no :5. .2. 28 33530
we um mm mm 2 mm ax; 25:9 - $052053. 55o 295
mm 2 S Q S 8 A5 2:29, S _o=§_o.§£o 29;
odd a a mm mm mm om _~ €85 3:3. 29:0,,—
ofia an 2 E 2:. a a: N: €35 8:2 2a:
A: Aouaugav 3:3 2:3 a: 32:30
8 Ba..."
«U «32 =c=o=< 33.395 {EEK-5 =o=u=< .8839..— Euoaam
Hm 2m 35— 880 2&2.on

 

658:0 3 233 .«o momma—83m can 8:3, 28:on ”2 28¢

50

900—35303 :5 Q8302: .:0:0:3 .023 $520 05 0: £80822 0.: 2 32.20:: 0: 00535 g 20:00: .3300 3:830 :5 3087.32 50: .0: vcoméoom
wEB: 33000. 2:05 a .0023 0.0:. :5 0.25.: £300 .2003 .2020: £020 .m=:m m: :03 00:03:00 0.300 =: :0 Sam 05 2 0.09m 2:00 8v 6000000.: 0: 0:
E05 :00. 0: .020 2 0.22:: :3 22:: 8:05 9:303 wESEwg—mv .wx cmnéom 5:: .m: can E0: 202:: wcsoa 00—5 :2: 000.: 0:: 20:80.20 wEfiEE 6:0
.9. 8m 5:: 32:0 $220. :02005 £0300 :0 5:058: 0:: 2 :0:E0:0 200-300 ”mm :0::0: 0:03 205. 5:03:03 .22% 0:: E A3 xmcom £0222 comwom
5:038: 305 “:0 :0308035. :5. :5 50 “08:0: :5. :0 :50 05 $2 2 Ban—0:00:05. 02: 2 :85. 3:. 62:5: :0: 328:0 2.2 020 3:30 2.2 .«0 83
05 2 02¢. Eh: =E0>O A3 ”35:02: 0 :02 05 wESE 20:02.0: £003.05 0,: 2 w:_>: 0:08:08 2:25: :0 .0983 .80: 0.: 2 02m 20:00:03 AS 62 88:
02m 20:00:05 :38 :0>0 3:02 we :05 .020 23 £00» 2 :05 03:0.» 20.2030: 0.: 2 0:022:88 :0 .025: 0.: 2 0:8 20:02.33. AS 60:00.60 8.322%
be .2 v 55:02:: E :0::0::0 >5 .2 m 60:00::0 b03003 >5 .2 N 60:00::0 ESE:— ba .2 _ 0: 75:0 .:0:00::0 0: .2 c 0: .0300 m. :0:00::m CV 3802

 

 

3 2: 2: g 8 3 22.0 06:80.0 8:
a,“ E z 2: 3 R 2: 22.0 E253 :o 28:. a 0.25 a:
c n v n 2 a :5: Be 822:8 2 852:0
0.: R 2: m: a mmm 2m 5 :5: 00:2: 9 823:0
- ca - - - h E 53 on: a: $30 2020
A2: E22028
3 mm b 2 0m vm 0053230 225200: 32000:
2.0825553:
3; m: a: 0: 2: 2: 2.0:: 2 2a 2.5.5:
.00... .00...
m0 :2 Z :0:0:< 30.0000...— :_:._0::m :0:0:< 38000.:— «8.0::m
Em Em «0:: 3000 2.92.22

 

A:0:0::::0UV 3:383:20 802005 :50 8.8: A: 030.:

51

 

 

_0_ a: $0 :0: E: 0.: 0. 30 600:: 0.00:0 0:000

0.:.0 0.: 0: 00 2: :. R 00 802000 02:02::

mm 00 00. 0: 00 S 5:280 020.00%

3 mm om m m 2‘ m0 5:830 200-300

- 0: m 0 m m N 0203:2000... 00:00:00.: 022

0.: 3 0: a 00 m0 :0 020. :5 000:: 08050 2:00:
. . A3 2:00:32:
:0 0\0V .05. :5: 0009:: 8:032:

E 0 0 0:- m. 2 02.0 005:0

0.:.0 0.: E 0.: R0. 000 000. 0.0:. .000 00002 0::
E05202:

0.:.0 0.: $00 :0 :5 2: E: :00 A000 :00. a: ~00 200 3:. 0.0000 000 :00: 2.: 0:00.50
A3 $023005 02: Eh: =0..0>O
0.:.0 0.0 0.0. :m 0.: 0.0. 3 5 2000:5020 00: 20:002.:
0.0 0 0m N: 3 mm 0:. :0 5 02.0 0:00 2000:0000:
a a E m m N. A0\0v 00200:: 0.0802
0.. I I E 0.. 0.: E 00:80:00 0002000:
:.0 2a SN :8 0.2 0.00 «.0: 20:00: 0200 .23 85:09.0
0.00 00: 0.0: 0.2. 0.0... 0:. 0:000 0002000: 00 00¢.
:0 mm 3 0.: fl 0: 90 0.2000

.00... .00...
:0 :2 Z 530:4: 320.000.:— aE._0:_:m :0_:0:< 300000...— «5.8::m
Em Hm :02 30.00 33.322

 

0052802020 50305 :50 8.8.2 “o: 030.:

52

8.50:0 3085 05 50 m0wfim 082:8 00 005006080008

08208 00 0:0 00 800 0005. .1 8080085 0008 00 80800 00800080 8050088000 0050 :05 03030: 80:0 3003 05 50
_0>0_ 008 00 80805008 0030800000 80308008 _00€0> 0 8800—50.“ 00090 :30“ 030 .805 0>0s 00 80000008 080m ... 3.80 Z

 

.8325...—
.30.;

X
X

0000:? 0080:? 0000.. «000800 000—002
00000 - 8050.0: 9.03005000m «am
- 00.00

 

$8005 :0_.0_.::00U .00_20>

8000005 a. 08:02 :0:0=< =00>aom Am

80:02 00:22 0% 00000005 =00E0m AN

0030005 0% 0.0500098250. 3:20:00; :00E0m 2
5.2.2002: 5.2.0 2.3.6 0.0.; 2.. 5

8000005 0% ~05 =00§0m Am

80000005 03,—. 00100—002053 00~__0_000wv 5— 000.50m AN
C0_0m0_0£s

30:02:50 _0_ 0 0.05 335303 0508: :8 52500 2
3.80.9? 8080.500...— m:_2_0 000208.395

800.605 0% 80000008 0.08-0804 c0050m Am

«80000008 208-0904 0% 00.000 00:30

00000005 0:0 80:0 “8:08.585 58:02 toaxm 000.55 C
"0889?

8080.800...— 80000005 00.:0t0-t00~m 0.000-093

<0:— <hmOU :—

828080... 0.2.0 3.5.5

 

4003038808 850:0 NEE—m 500m 05 8 0.053000% 0080800000 .0030?» “a 0305:

53

Table 5: MarketingChannel Selection Estimation Results (WESML)

 

 

Number of obs = 145
Wald chi2 (28) = 7724.76 Prob > chi2 = 0.0000
Log pseudo-likelihood = -51.928071 Pseudo R2 = 0.6061
Covariate CSU-Hortifruti La Colonia
Producer is female -0.155 1.662
(0.88) (1.59)
Age of farmer (years) 0.037 -0.078*
(0.03) (0.06)
Experience in tomato production (years) -0.093** 0.097
(0.04) (0.06)
Producer education (years of schooling) 0.155* -0.027
(0.09) (0.15)
Household size (members) 0.183 0.057
(0.24) (0.18)
Dependency ratio 0.362 0.758
(1.26) (1.93)
Producers has means of transport (dummy) 0.626 -1.069
(0.71) (1.13)
Livestock (head) 0.009 0.035
(0.10) (0.04)
Producer has electricity (dummy) -0.67 1.929
(0.99) (1.91)
Distance to asphalted road (kms) 0129*“ 0080*
(0.04) (0.04)
Overall farm size (ha) -0.064 0.044
(0.05) (0.04)

 

Notes: 1. Robust standard errors in parentheses. 2. Parameter * significant at 10%; **
significant at 5%; *** significant at 1%. 3. Base Category is traditional marketing chain.

54

Table 6: Second stage: Production Function Estimation Results

 

 

Covariate CSU-Hortifruti La Colonia Traditional
Labor (family and hired days) 0.621 ** 2084*“ 0609*”
(0.33) (0.85) (0.26)
Fertilizers (U S$/farm) 0.203* 0.423*** 0034*
(0.12) (0.18) (0.21)
Foliars (USS/farm) 0.124 0.157 -0.072
(0.26) (0.27) (O. 12)
Insecticides (USS/farm) 0.699* 0.31 0.1 15
(0.38) (0.25) (0.15)
Herbicides (USS/farm) 0.135 0.018 0.044
(0.07) (0.05) (0.04)
Fungicides (USS/farm) 0.426* 0.322 0.04
(0.20) (0.15) (0.06)
Seeds (USS/farm) 0.095* 0.084 0.062
(0.02) (0.10) (0.10)
Total tomato area (ha) 0.134* 1.189* 0.442***
(0.05) (0.69) (0.12)
Irrigation (% of tomato area 0.126 0.092*** 0.126***
irrigated)
(0.25) (0.08) (0.06)
Tractor use (1=yes) 0243* 0.016 0.054
(0.09) (0.57) (0.10)
Mills] 0.056 0.03 -0.012
(0.04) (0.02) (0.02)
Mills2 0.018 -0.005 0.001
(0.02) (0.02) (0.02)
Constant 1.686“ 1851*" 1.052*
(0.73) (0.34) (0.62)
Observations 43 26 76
R-squared 0.73 0.72 0.63

 

Notes: 1. Robust standard errors in parentheses. 2. Parameter * significant at 10%; **
significant at 5%; *** significant at 1%.

55

X 7 8003005 Q8102 :0..0=< c0050.. Am
000.002 :0..0=< Q 0080008 80.0 =08m :00E0m 3
X 00000008 80.: =08m Q 80:000.. 3:00.005. :00E0m A.

$2803.09... 52.0 5.00.5 0...:- 0... =—

X 8000005 Q .0. :0050m Am

X 00000005 Q €208.00? 000:0.0003 .0. :00300m Am

X €203.00?
03:283. 5: 0 002° 08.08.83 05080 000 52200 c

"08000.? 8080.800...— m:.0..0 000—0080055m

X 8000005 Q .0095. .0:0.w0m :00E0m Am

X 0009:. .8083. Q 8000,0008 0.000-080; :00505 .m

X 080000008 2000-0804 Q 0000.0 00:30 0030005
0:0 8.0000001080003 0:000m 800.002 8095 0003.05 A.

”080.0%

8080800.:— 8080005 00.00.009.095 0.000-090.—

«504935;.

X

Eczaox: 00:0...0 00:0...< 00.000 000.5000 30.00.).
.00.t0> 0000.. - 00.00.00— 0003005025 000w 80.00080... 58.0 ban-5
- 00.00

 

380:5 8.0.8.0800 .00_t0>

 

0808.0 5.59% 500m 0... 8 m0.w0.0bm :000800000 .00.t0> no 030-.-

56

.000.-.Ew.m*** .Xm .0 .000me... I. .200. .0 .000...00w.m * 0000500000 0. 00000 0.00005 "00.07.

0\0. .0

 

 

00.0 00.0 80 8.0 0.0 M00 .00- 00.0- 000 $0520....
80 00.0 0.0 000- 00.0 000 0.0 .00- :0 0000 000%.... 0.0.00..-
0m. .00 0... N00 000 .0.. 00.0 mm. 000 0000 0.00.00-
00.00.0000 00.00-00:
N00 ~m0 0 . .0- 00. .- 0N0 .00- ~00- .00 .00- 0.000. 0000.002
000 3.0- m00- 0.... 00 .0- 000- 000 000- 000- 0003
00.0 . .0 . .0 0N0 2.0 0.0 $0 0.0 ~00 0.00.0. 00.00
00.000020 00.0..
000.000
.000. 0.005 000.000 ~..00..m 000.000 0 0.00.5
02.0. 000.00.. 2.0.00 .00... 0000.00. 2.0.00 .00... 00.00: 0.0.00
-007. 02.0.. 0.000... 0000.007. 02.0. 0.0.00... 02.0.-..07. 02.0. 0.0.00... 00.00.00... 00.00.0000
.000...000... 0.00.00 01. 0000.00:

 

00000000000. 3.000.. 00..00...mm. 0000.00,... 0000.00 .000. 80.00.0080. 000 0.000m 000.00 “0 0.00..-

7
5

 

 

0. 0. , .0. 22028.0
00... 0.... .0... 0.... N0... .0... 8... 3.... 2... N...
88. 6:. 8.8.. .000... 88. .83... .8... c... .08...

a..- a..- 8...- .0...- o.. 3.... 3.- m: .00: . .5088
0... .0 .0... .80 .8. 8.9.. .2. 00 a...

0 a- as. .. .0 .m- N... N... v 0... .0205
8... a. 0.0... 0.0.. .2. .03... 3.. 0.. a...
.00- .0. 3...- v..- 2- 0.0. m .2- 3.- 00:0:
0.... 5.. .28. .8... 2... 8.2. 2 . .. 0... .3...

a a. 0.0. 0.. R. 2:00.. :00.- Nm- :8... 00520..
G... .0... 030. .00... .8: .000... 80.. 0.... 0......

8 v. .8. m...- N :32... E :- :02. .0 3.. 03.00... 9250
.00. 0:. .8... 0.0.. .2. :8... E. 0... 002.

1:0... 2:03 58... 2...... 2...... :3... 1:80 8. .00.... 0.... 80 9.050
.0... .00. 0.8.. an... .0... 00.2. G... .2. 0......

.3000...

MR :0. :0.- ......00. 3 _ a..- :00. ..Nm :8...- 02... 3.2-82
0.5.. 3.. .080. .30.. G... .0008. .20. 0.0. .00....

*

00hw©w

2. .3.- .fi- 0.: 2..- $0...- .0 1.:- - 9.0000. 00.3
a... a... .88.. at... 6:... 8.3.. a... a... 0.8....

0... 3. 80: .3. $0 .80.. 8: n: .2... $00.00. 32.9 80..

000.000
.00... 0.005 000.000 0.0005 000.000 0 0.005
02.0. 000.000 .00.00 .00... 000.000 00.00 .000. 00.000 .00.00
-00 7. 02.0.. 0.00.0... 02.0.-002 02.01. 0.0.00... 02.0.-002 02.01. 0.0.00... 00.0.0.0... 080.000....

0050...... 00.200 3 0:000...

 

0000.05 8.000.. 000000000 0000...... 000.000 3.00. 80.00.0080. 000 0.00.5 .0080 ”m 0.00...

58

CHAPTER 3: ESSAY ON CATTLE PRODUCERS’ PARTICIPATION IN
MARKET CHANNELS IN CENTRAL AMERICA: SUPERMARKETS,

PROCESSORS, AND AUCTIONS

Abstract
This paper focuses on the determinants of the participation of cattle producers in the
supermarket channel, export processor channel, and traditional auction channel. It begins
with an analysis of the market channels using qualitative data from 50 interviews of
retailers, processors, auction market managers, and other key informants in Costa Rica
and Nicaragua, two widely differing cases. It then analyzes patterns and supplies of

producers by channel using farm level data from 300 farms in the two countries.

Key words — Multiple market determinants, supermarkets, Central America, Costa Rica,

Nicaragua, cattle producers, supply response.

59

Cattle Producers’ Participation in Market Channels in Central America:

Supermarkets, Processors, and Auctions

1. INTRODUCTION

Today in development circles interest is high in whether and how farmers can
gain access to markets for high-value products such as livestock, dairy, or fruits and
vegetables. Access to such markets can open opportunities in dynamic urban markets and
diversify production away from basic grains. Research in the past several decades on

farmers’ participation in markets, including livestock markets, has followed several paths.

First, studies have been done on producers’ market participation as net sellers or
buyers, versus subsistence production or autarky only. Examples of such studies include
Goetz (1992), Key et al. (2000), Holloway et al. (2004), and Bellemare and Barrett
(2004). This research contributes to understanding the complexities of the factors moving
producers from a subsistence orientation to commercial production. It has been pointed
out by Pingali and Rosegrant (1995) that rural developers have a challenging task in the

transition from subsistence to commercial production systems.

Second, given participation in cattle markets, a number of studies have focused on
supply response models that incorporate the channel choice between export and domestic
markets (Jarvis, 1974; Ospina & Shumway, 1979; Spreen et al., 1980; Paarsch, 1985;

Rosen et al., 1994; and Aadland & Bailey, 2001).

60

However, the literature has not given much attention to the determinants of
producers’ multi-market channel choices, particularly in domestic markets. An exception
is Hobbs (1997), who analyzed the determinants of auction market channel versus

processor market channel using a sample choice of cattle producers in the UK.

The need for additional research, in particular in developing countries where
markets are transforming quickly, on producers’ multi-market decisions arises from the
emergence of new market channels in domestic markets with different procurement
structures that bring with them additional market requirements (such as quality standards
and consistency). The geographical context for this paper provides an illustration. Since
market liberalization, and especially foreign direct investment (F DI) liberalization,
occurred mainly in the 19903, supermarkets have become very important buyers of food
products from farmers in Latin America. Supermarkets have increased their share in
urban food markets from 10-20% in 1992 to 50-60% in 2002. Moreover, leading
supermarket chains have adopted modern procurement systems shifting steadily toward
direct procurement from producers, creating market conditions different from the
traditional procurement methods of buying indirectly from farmers via public wholesale

markets, intermediaries and processors (Reardon and Berdegué, 2002).

In addition, competition among domestic supermarket chains and the desire to
differentiate themselves from their competitors in traditional markets, and from each
other, as well as domestic and international food safety crises (such as mad cow disease)
and internationalization of food systems, have given birth to the dominance of private
standards of quality and also safety. The emergence of bigger buyers such as

supermarkets and export-oriented processors to capture the gains from market

61

differentiation have introduced changes in conditions of agro-food production (Henson

and Reardon, 2005; Codron, Giraud-Héraud and Soler, 2005).

Hence, institutional change and market channel differentiation translate into
different requirements for farm technology and transaction practices. Some farmers are
able, because of greater human, social, organizational, and financial capital resources, to
meet the requirements of the higher-value market channels, and some are not, and may be

eventually left behind as markets transform.

The above-mentioned changes in certain markets and regions take the research
agenda from whether farmers participate in “the” market, to in which market channel
farmers’ participate and with what entry requirements and with what implications for

their net incomes.

In this paper we aim to contribute to the multi-market channel participation
literature by exploring the factors affecting producers’ market channel participation by
including the supermarket market channel with other market channel alternatives such as
the processing market channel (with an export orientation) and the traditional auction
market channel. We compare producers’ supply determinants by market channel. The
study covers Costa Rica and Nicaragua, countries of similar population (small countries
of about 4 million each) but very different per capita incomes (Costa Rica 2000 GDP per
capita of US$ 4,290, while Nicaragua’s was only $600), and different stages of beef

industry and supermarket sector development.

62

The paper starts by describing the cattle supply chains in Central America. We
first present a conceptual framework, then we discuss, for Nicaragua and Costa Rica
respectively, the methods used to collect our qualitative data on the supply chains,
followed by an analysis of the processing industry procurement systems, the leading
supermarket chains procurement systems, auction markets, and a discussion of the
characteristics of market channels cattle producers face. The following section treats
producer characteristics by market channel, and presents the descriptive and regression
results. We use data from a farm household survey with a sample of 300 farms in Costa
Rica and Nicaragua to estimate the market choice selection and supply decision. The final

section presents conclusions and implications.

2. CATTLE SUPPLY CHAIN COORDINATION MECHANISMS

To set the context for what markets producers face in Central America, here we
describe recent changes in the livestock markets. We first present the method used to
collect the data to analyze the supply chains. Thereafter, in order, we present a conceptual
framework for the analysis of the supply chains, and then the qualitative analysis of beef
supply chains in Nicaragua and Costa Rica. There are three supply chains : those
supplying large-scale export oriented processors; supermarket chains; and auction

markets.

(a) Data for the Qualitative Analysis of Supply Chains
We collected qualitative data on the three supply chains (the terms “supply

chains” and “market channels” are used interchangeably) from 50 in-depth in-person

semi-structured interviews with retailers and wholesalers in the modern and

63

traditional channels, NGOs, international research institutes, group leaders and
members of groups of farmers, community leaders, technical assistance providers

(N 605 and extensionists) and other key informants. The interviews focused on the
relations between the agents in the supply chains: (1) what'agreements (oral or
written) exist, and what are their “legislative functions” such as transaction product
attributes and transaction requirements (product quality, price, and consistency); (2)
what were the reasons for the implemented agreements; (3) who participated in
designing them; (4) who implements and enforces the agreements, such as
administrative functions including cattle selection, production planning, price
negotiation, delivery, reception, and distribution of payments; (5) what sanctions and
incentives exist in the agreements; (6) what are the advantages and disadvantages of
current and alternative coordination strategies in terms of costs, consistency, and
conflicts; (6) how have the answers to these questions differed between now and five
years ago. We compared and triangulated information over actors to form the

composite image presented below.

(b) Conceptual Framework for Supply Chain Analysis: The Vertical Coordination
Continuum

We use the Peterson, Wysocki and Harsh (PWH) model as the analytical
framework to analyze the supply chains in the study (Peterson et. a1, 2001). The model
aims to be a decision making tool that helps understanding of how and why the agents
along the supply chains (supermarkets, wholesalers, producers), choose certain strategies

from a vertical coordination continuum. Several points should be emphasized.

64

First, “vertical coordination” is the configuration and control of the stages of
production and marketing systems. For a specific product such as beef, this means that
vertical coordination defines the nature of the specifications by which the cattle are
produced, processed, as well as who is involved and under 'what conditions in marketing
the cattle at each link in the supply chain until they reach the final consumer. As there are
potentially many stages and agents involved in production, processing and marketing of a
product such as beef, there are also many relations under which the transactions could
take place. Hence, part of the vertical coordination process are the institutional
arrangements governing those relations. There are some important related questions such
as who defines the institutional arrangements and the penalties for not complying with
them. An example of the “who” is “who defines the minimum weight and age of cattle

that ought to be marketed?”

Second, “continuum” here is the set of choices that a given agent in a supply
chain has to vertically coordinate its procurement systems (for example, procuring live
cattle or processed beef). There are many potential ways to vertically coordinate the
procurement of cattle (or processed beef) so some criteria to categorize the alternatives
are useful. The PWH model defines the vertical coordination continuum as an array of
five groups of strategies: spot markets, specification contracts, relation-based alliances,
equity-based alliances, and vertical integration. As firms move from spot markets
towards vertical integration they increase the level of control of the supply chain,

assuring proper coordination.

Third, higher levels of control might be desirable when coordination errors are

costly. Coordination errors happen, for example, when there is not enough supply or

65

when the quality is not right. For example if a supermarket runs out of its stock of beef or
if the beef lacks tenderness, consumers may shop somewhere else. Coordination errors of
not having enough supplies or the proper attributes may translate into losses in sales.
Coordination errors happen, according to Williamson (1973), because of intentional
opportunism of economic actors or unintentionally due to bounded rationality of

estimating production plans (under or overestimation).

Fourth, as one moves from the spot market to hybrid coordination strategies to
vertical integration the nature of control changes. In spot markets the intensity of control
is low, the coordination mechanism is based mainly on price discovery, and the decision
is whether or not to engage in the transaction. There is no establishment of coordination
mechanisms beyond the transaction itself. This type of coordination mechanism is
equivalent to auction markets. The next level of control is represented by the
specification contract, which is a coordination mechanism in which the agents engage in
a transaction by specifying a legal contract. The established specifications in the contract
are negotiated carefully, setting incentives and deliverables. Part of the contract might
include the use of third parties to enforce the conditions of the contract. The third strategy
is called the relation-based alliance, and its name invokes an established “relation” in that
benefits and risk are shared as well as mutual objectives identified. Parties work closely
together to resolve conflicts internally and mutual control is executed. The fourth position
is the equity—based alliance, which involves some level of equity capital such as joint
ventures. The main characteristic is the formation of a third entity distinct from the
parties that has a formal organization structure. There is financial commitment and the

rights and responsibilities are clearer as compared to the relation-based alliance. At the

66

end of the continuum is the vertical integration vertical coordination mechanism. In this
strategy the two parties become one organization (by a merger or acquisition) and it has
total control over the transaction. Clear policies and procedures are set for the new single
organization and the notion of centralization is more important than the notion of single

organization as the key is to centralize control for coordination purposes.

Note that coordination strategies might differ at different levels of a supply chain.
Consider a beef supply chain where there are four agents (supermarket, wholesaler,
processor, and producer) and three relations (at the linkage points). On one hand, the
supermarket, the wholesaler and the processor may use a relation-based alliance
coordination mechanism. On the other, the wholesaler and the producer might use a spot
market strategy. Importantly, the former strategy of an alliance is shared, while the latter
is not but is the result of separate and unrelated decisions. Strategies might differ because

of the likelihood and costliness of coordination errors of each relation.

Fifth, agents in supply chains choose their vertical coordination strategy along the
continuum based on both the coordination error costs and the operating cost of a given

strategy.

Measuring coordination error costs may be challenging, but PWH suggest two criteria to
assess the costliness of a coordination error. On the one hand “asset specificity” is the
extent to which an asset can be utilized in alternative uses. An example of this can be a
cattle processing plant that has a specific type of sanitary certification but that can not be
used to process other types of meats. Engaging in a relation with this processor may be
the only way of having the beef process under the specified sanitary conditions. On the

other hand “complementarity” is the notion that the output from combining two activities

67

is greater than the output of each activity separately. An example of this can be that a
retail store may offer organic beef only if a group of producers agree to follow organic
production specifications. A retail store may not be able to offer organic production
unless producers use only organic approved inputs to fertilize their prairies and follow
other guidelines for organic production. Animal welfare ensuring the animals have been
grown under certain conditions also represents another example. The costliness of

coordination errors rises as both the asset specificity and complementarity increase.

Sixth, besides evaluating the costs associated with each coordination strategy, an
agent must consider other conditions. Implementing a strategy includes conditions such
as capital availability, where spot markets are associated with nearly no capital
requirements while vertically integrating may entail significant capital outlays.l4
Existence of a compatible partner may be important, to help to ensure mutual interests are
achieved. Control competence entails that the partners have the necessary skills such as
those needed for the type of control required in a strategy. Finally, anagent has to judge

the risk/return tradeoff of shifting from one alternative to another.

(c) Beef supply chains (market channels) in Nicaragua

We start by presenting the structure and procurement systems of the processing
industry. Note that in Nicaragua, the processing industry is composed of many small
processors and a few recently established large-scale processors; the small processors and
the large processors differ greatly in food safety aspects. Then we discuss the
supermarket chains’ procurement systems. Finally we discuss the auction market as a

traditional channel.

68

Large-scale Export-Oriented Processors ’ Procurement Systems. First, the
Nicaraguan beef processing industry has recently consolidated, and a few large-scale
processors have displaced many small processors. By 2004 there were three large-scale
processors left standing, with 60% of the market: MACESA, Nuevo Carnic, and San
Martin. Each has a processing capacity of over 300 head per day on average. Contrast
that with the situation in 1994, when there were nine large-scale processors
(slaughterhouses) with a processing capacity of 150 head per day. On the other hand,
according to Pomareda et al. (1997), during the mid 19905 there were 300 small
processors (slaughterhouses) in the rural areas and 600 municipal small processors in the
urban areas in Nicaragua, each processing only about 15-20 head a day on the
intermittent days they operated. In the late 19905, the large-scale processors increased
their cattle volume substantially in spite of being fewer, and thus gained market share at
the expense of the many small slaughterhouses which did not increase their volume
(IICA-PROVIA, 2002). Moreover, several of the rural and municipal slaughterhouses

closed during the late 19903 and early 20008.

The higher efficiency of modern technologies used in the large-scale plants and
changes in the international markets explain the consolidation, including the
displacement of the smaller processors. Note that the three large-scale processors are
HACCP- and USDA- certified, which address international markets’ food safety
standards. The animals’ health is inspected 12 hours before being processed and also after
being processed, by veterinarians from the Public Health Ministry and the Agriculture
Ministry. Sick animals are not allowed to be processed and animal carcasses are

inspected for “muscular cisticercosis” and internal signals of infection, parasites, and

69

other diseases. Also, lab testing of samples is performed to know if there are implants and
chemical residues above certain maximum levels. These safety checks are done because
of the export orientation of bigger plants, while they are not done in the small plants
focused on the local market. According to Campbell (2000) who visited two of these
small processors and reports that they operate without any refiigeration, with inspection
of the cattle, and contamination from previous slaughter is visible and represents a
concern for public health. Pomareda et al. (1997) report that with the exception of two
operators (Brasiles and Proincasa), all the small processors produce “carne caliente”
(“warm meat”), a term used for the marketed beef that has not being refrigerated after the

animals have been slaughtered.

Moreover, domestic market changes led by changes in supermarket chains’
procurement systems accelerated the displacement of the smaller rural municipal
processors. The two leading supermarket chains sell only cattle that have been processed
in the bigger, HACCP-certified plants. They do not buy from the small processors
because of the lack of food safety standards, lack of refrigeration, and because the quality
and volume of cattle required by the chains is several times higher than the small plants’
processing capacity. As supermarkets gain market share in domestic retail markets, the

bigger processors benefit from this growth.

Third, the large-scale export-oriented processors’ cattle procurement systems rely on
a list of buyers in the production regions to get their cattle. The buyers are assigned
regions from where to buy their cattle. There are no minimum quality standards but there
are differences in prices paid by animal sex, weight, and age. For example, the leading

processor San Martin had a group of buyers that were part of the company, but today

70

work on their own supplying the slaughter plant. In addition in times of shortfalls, San
Martin uses 60 external intermediaries to buy animals. Only a few producers sell directly

to the processors.

Supermarket chains ’ procurement systems. The share of supermarkets in food retail
in Nicaragua has reached 15-20% today, and rising (Berdegué et al., 2005). Today the
Nicaraguan supermarket sector is made up of two supermarket chains (with 37 stores)
with about two-thirds of the market, and 23 independent supermarkets. The supermarket
sector’s leader is Corporacién de Supermercados Unidos (CSU), a chain with about 65%
of the market (CSU is headquartered in Costa Rica, hence it is a regional multinational
also present in Honduras). In second-place (with 20% of the market) is the supermarket
chain La Colonia, the largest locally owned chain. They compete with hundreds of

traditional retailers, such as wet markets in each city, and small meat stores.

The two supermarket chains operating in Nicaragua do not source cattle in the same
ways. On the one hand, “La Colonia” does not buy live-cattle either from auction markets
or from producers. La Colonia buys all its carcasses and beef cuts only from the three
large-scale, certified, and export-oriented processors rather than from small processing
plants. By working exclusively with the large-scale processors, the chain achieves higher
food safety standards as compared to those found in the small scale processors. The chain
centralizes its purchases of beef; however, because it has neither a distribution center nor
trucks, it requires the processors to deliver the beef carcasses to each store. The chain
negotiates quotas with each of the three large-scale processors mainly based on price and
payment conditions. Recently the chain shifted from buying mainly from the large-scale

processor “San Martin” to buying mainly from the large-scale processor “MACESA”,

71

because of more attractive payment conditions. The latter formed an alliance with the
local bank “LAFISE—Bancentro” to offer to La Colonia an extension in the payment
period from seven days to 30 days.15 This latter point is particularly important, as small
processors generally have a hard time weathering the long payment periods of
supermarkets, but here with MACESA is a large processor who allies with a major bank

to handle this financial hurdle.

On the other hand, the leading chain CSU buys all its cattle from “Industrias Carnicas
Integradas” (“Integrated Meat Industries”) (ICI). CSU and ICI are based in Costa Rica
and are part of the same holding company. CSU started to work with ICI because it could
monitor easily and get consistent supplies of beef, and thus meet CSU’s safety and
quality standards. That would let CSU focus on retail expansion grth strategies while
ICI reduces beef procurement costs and it would be able, if needed, to supply CSU stores

in other countries in the region.

ICI in turn has invested in trucks and a distribution center in Managua, the capital city
of Nicaragua, and from there it coordinates its deliveries to CSU stores. ICI established
contracts to use the plant of one of the three large-scale export oriented processors (San
Martin), to process all the animals ICI procures. ICI does not buy cuts or carcasses from
the processor but instead pays for the slaughtering service. ICI and San Martin have
implemented some changes in the way beef is marketed in the domestic market, replacing
the typical half carcasses by beef boxes. San Martin in turn has collaborated in the
implementation of such changes in the domestic market. ICI then minimizes the beef

processing requirements both at its distribution centers and at C SU stores.

72

ICI in turn buys live-cattle directly from 130 producers listed in their preferred
supplier program, an implicit contract relationship. Only in cases of shortage do they
make up the shortfall with purchases of cuts from the three large-scale processors. In
2004 they also had a herd of 1200 animals managed by a farmer near the plant, so they
had some own production to supplement purchased supply when there are shortfalls in

the latter.

The CSU-ICI buyers have developed their own procurement system in which
producers are enlisted based on their commitment to meet quality standards and
transaction requirements which are different from those of the export-oriented processors.
For example, ICI buys animals the carcasses of which should weigh a minimum of 212
Kg, which is 15% higher than what is required by the export processors. Also, ICI buys
only steers which is the animal category with the highest quality - while the export
processors buy other animal categories of lower quality such as cull cows. Hence ICI has
higher quality requirements than the export-oriented processors. Producers in the
preferred supplier system must sell other animal categories, considered of lower quality,

such as cull cows, to alternative market channels.

Note that ICI’s sales are not limited to CSU domestic stores but it also exports to
supermarkets of CSU in Costa Rica and has supply contracts with La Fragua (the other
leading chain in Central America, based in Guatemala, with which CSU is in a joint
venture) in Honduras (Dispensa Familiar, Paiz), El Salvador (Dispensa San Juan) and
Guatemala (Paiz and HiperPaiz). An example of ICI’s regional procurement and
marketing capacities is that during 2003 only 25% of what ICI bought in Nicaragua was

sold in Nicaragua, and 75% was sold to its international subsidiaries and foreign

73

customers in the region. These transactions are all taking place within the multinational
regional chain the Central American Retail Holding Company, CARHCO (CSU, La
Fragua, and until 2005, global retailer Ahold, and now, Wal-mart instead of Ahold). ICI
in Nicaragua actually “exports” (to stores in the same joint venture regional chain) more
animals than it provides to the CSU stores in Nicaragua itself: for example, while ICI
Nicaragua exports 2800 animal carcasses to El Salvador it sells 1200 per month to its

Nicaraguan CSU stores.

Auction markets as an alternative market channel. First, consider that both ICI and
the large-scale export-oriented processors do not rely on cattle auction markets to buy
their animals. They mentioned that sanitary conditions are not optimal; hence animals
may get sick or hurt when bought from there. A second reason is that they often had
difficulties because the quantities and quality of animals sold fluctuated too much, and
they would have to find alternative cattle sources because of their high volume
requirements. A third reason is that the auction market had an extra-cost that requires
paying seller and buyer commissions to the auction management equivalent to 5 to 10%
of the value of the transactions. Note that there was only one auction market in operation

in Nicaragua during our study period.

Second, what is traded in the Nicaraguan auction market is subsequently processed in
the small processors and sold in traditional retailers. This is further corroborated by the
fact that the large-scale processors do not provide the slaughtering service to any other
agents or firms than ICI, which eliminates the possibility that the buyers at the auction
market process their animals at the large-scale processors. On the other hand this does not

necessarily mean that everything processed in the small plants is traded at the auction

74

market. Indeed there is likely to be an important volume of cattle bought from the farms,
then processed in the small plants and sold to traditional retailers. Nevertheless, what is
traded in the auction market is then in the domain of the small plants and traditional

retailers.

Third, it must be that the reasons why the large-scale processors and ICI do not buy
from the auction market do not stop the buyers in the auction market sourcing from there,
then take it to a small processing plant and then to a traditional retailer. Lower sanitary
conditions, higher fluctuations in quality and volume are tolerated or at least not an
impediment to the buyers that go to the auction market, and the extra cost paid
(commission) may be trade-off for the costs associated with searching costs with getting

their volume requirements directly from some farms.

(d) Beef supply chains (market channels) in Costa Rica

Compared to Nicaragua, in Costa Rica, the processing industry is more concentrated,
traditional retailers’ process their cattle under safer sanitary conditions, there are more
supermarket chains operating with a higher share of urban food markets, and there are
many auction markets. Hereafter, we present the structure and procurement systems of
the processing industry, procurement systems of supermarkets chains and auctions

markets.

Large-scale Export-Oriented Processors ’ Procurement Systems. First, the
processing industry is much more concentrated than in Nicaragua. In the beef processing
industry there are three major processors (Coopemontecillos, Arreo SA. and Mataderos

del Valle S.A.), which process 80% of the total national cattle slaughtered in Costa Rica.

75

The first two have HACCP and USDA certifications, which are standards, designed for
food safety procedures (such as monitoring for the presence of salmonella and E. coli in
processed beef) and implemented mainly to target international market food safety

concerns.

Second, the major processors started focused on export markets but have shifted
toward the domestic market over time. Costa Rica exported 50% of its production during
the 1960’s and 1970’s; that share fell to less than 10% in the 20003. The decline in
exports may be linked to the decline in the Costa Rican national female cattle stock,
which in turn might be associated with negative future expectations and decline of the
competitiveness of the sector relative to the Nicaraguan cattle sub-sector. Exports are not
that important for the Costa Rican industry now in terms of volume, but when processing
carcasses some portions may easily be sold in the international market (i.e. less valuable
lean cuts transformed in the US into hamburgers by blending them with US fatty cuts).
Some cuts are difficult to sell in the domestic market and the processOrs may end up with
high quantities of inventories. Costa Rica’s two major processors (Arreo and Montecillos)
consider it very strategic to maintain the possibilities of selling to both the domestic and

international markets.

Third, beside these three national-level plants, there are three regional plants and 12
smaller rural plants, which in the 19703 were more numerous (Pomareda, 2004) —- hence
more like Nicaragua at present. The smaller plants process less than 1,000 head per
month and do not have the proper sanitary and hygienic conditions (Perez, 2004). The
decline in rural plants seems to be a consequence of changes in food safety perceptions

by consumers, as well as enforcement of safety standards by the government (which is

76

more capable of enforcing such standards than in Nicaragua). These food safety forces
drove changes in both supermarket chains’ and small butcher shops’ procurement
systems: because of the low hygienic standards of the small processors they shifted to the
larger processors. Moreover, economies of scale gave the larger plants a competitive edge

over the smaller plants.

Fourth, the three major processors buy their cattle from producers, with the
processors’ assigning regional buyers in production zones. The processors use similar
standards to those recently developed by the Costa Rican Livestock Association
(CORFOGA). Processors link the standards to economic incentives to reward producers
for higher quality animals, which are younger and heavier animals, animals with reduced
fat content and good muscle formation. They do not establish a minimum weight
requirement nor only buy a single animal category, which lets a given producer sell all
his/her animal categories to this market channel. Processors also require the animals to be

delivered free of diseases and parasites.

Supermarket chains ’ procurement systems. Note first that in Costa Rica the
supermarkets’ share in urban food markets was 50% in 2002 (Alvarado & Charmel, 2002)
while their participation in beef retail markets was 35% (CORFOGA, 2002). The
supermarket chains face fierce competition from 1,389 butcher shops (traditional retailers
for beet). However, the supermarket growth has been significant in the past decade as
shown by the increase from 113 stores in 1990 to over 317 stores in 2003 (Alvarado,
2003; Perez, 2004). There are butcher shops in the urban open markets and in urban
residential areas. The food safety of butchers’ beef is similar to that of the supermarkets,

because, in contrast with Nicaragua, in Costa Rica the gap between processor’s food

77

safety used by the supermarket sector and the traditional retailers is smaller because the
processing industry is highly concentrated and there are fewer smaller operators. Also, in
Costa Rica butchers can take their animals to the large-scale processing plant and pay a
fee for the slaughtering service, which is not the business practice in Nicaragua as we

mentioned before (with the exception of ICI).

Second, supermarket chains have two types of procurement systems. In the second
tier, supermarket chains such as MEGASUPER, Supercompro, Perife'ricos, and
PriceSmart rely entirely on their relations with three major processors to buy their beef
cuts and carcasses. In this procurement system, supermarket chains and the processors
work together to define product standards and transaction requirements. Then the
processors are free to design and govern as they see fit their cattle procurement systems
and their relations with producers. The supermarket chains no not engage in direct
relations or transactions with producers, and they rather use, de facto and indirectly, the
standards imposed by the processors on the producers. This is equivalent to the La

Colonia supermarket chain we presented in the Nicaragua section.

Third, in the first tier, supermarket chains such as Automercados and CSU have
developed procurement systems in which they buy the cattle directly from producers.
Automercados has a preferred supplier system of 40 producers that supply all its cattle
requirements. This supermarket chain also owns a farm that is used to leverage their
negotiation position against producers and it is also used to supply the cattle in times
when competition for cattle is high. Interestingly, the chain has an internal department
that handles cattle transactions with their producers. The chain has contracts with two of

the major processors to process the cattle and deliver beef cuts to its seven stores. The

78

chain started to buy directly from producers because they wanted to ensure they were
buying high quality beef (steers) instead of low quality (cull cows) and the best way for
them to do that was to buy their animals. However, this high-end chain now buys more
beef cuts than live animals, because their further processing (in the store) capabilities are
very small. By buying specific cuts, they avoid inventories of unsold cuts (they order

what they sell).

CSU relies on its dedicated wholesaler Industrias Carnicas lntegradas (ICI) to get its
beef cut supplies. Recall that ICI is part of the same holding company that owns CSU.
ICI buys its cattle supplies from approximately 53 producers that are in their preferred
supplier list. ICI maintains a relation under a contract with two of the major processors

(CoopeMontecillos and Arreo) to slaughter the cattle.

CSU-ICI buys cattle because they are able to maintain the required carcass quality
and also secure their needs. In the past, and depending on the seasonal supply, the
processing industry did not maintain standards on the carcasses they sold to CSU-ICI,

and the same happened with certain valuable cuts (tenderloin).

CSU-lCI’s use volume of cattle is very large and the major processors compete with
each other to get the contract to slaughter the cattle for ICI. In the contract, ICI’s
producers send the cattle to the two major processors directly and ICI will get back from
the processors only the beef cuts and not the byproducts. Interestingly, it is the processors
that end up paying CSU because the value of byproducts is higher than the value of the
slaughtering service. If ICI is facing some cattle shortage situation or need to leverage its
negotiation position with the major processors, it requests its Nicaraguan subsidiary to

send animal carcasses.

79

CSU and ICI outsource the slaughtering because it does not give any added value
(only to maintain food safety achieved by the export standards). Slaughter is
"subsidized" by the need to attract a large volume of cattle. It is cheaper to outsource than
to build (and operate) a slaughterhouse. On the other hand, the slaughterhouses need
large volumes for the use of subproducts like bone- and blood flour, skins and hides. That
is why they offer services to butchers and ICI. The supermarkets use certified plants for

hygiene and quality reasons.

Also, in Costa Rica, ICI, besides having distribution centers, also has a processing
plant in Alajuela (near the capital, San Jose) where it further processes the beef by cutting,
marinating, and processing others products that are sold under four “private label” brands

for the CSU chain (La Hacienda, Lonja, Suli, and Sabemas).

As with the ICI in Nicaragua, ICI Costa Rica is also involved in CARHCO’s regional
procurement system -— it delivers its products to CSU stores in Costa Rica and also sells to
La Fragua stores in El Salvador and supermarkets in Nicaragua, Honduras, and
Guatemala. Note that lCI’s plant is HACCP certified and also has the Costa Rican green
ecological seal “Sello Verde de Bandera Ecologica” because of its environmental friendly

technologies such as waste water treatment.

Auction markets as an alternative market channel. First, note that there are many
more auction markets in Costa Rica than in Nicaragua. Interestingly, while at the
beginning of the 19803 Costa Rica had nearly no auction markets, currently there are 16
auction markets in different regions of the country. Also, the nature of the transactions in
the auction markets has evolved over time. The Costa Rican Govemment’s decree 26.920

of 1998 regulates the operations in auction markets establishing procedures such as

80

minimum health sanitary conditions, public cattle weight measurement and price
transparency (SIDE, 2004). Auction markets replaced a non-transparent system of
intermediaries. There are still a lot of intermediaries in the system but the auction system
skyrocketed because it satisfies the needs of the small prOducers who sell each on average
three animals in the auction markets and it gives price transparency and immediate
payment to producers. Producers sell mainly heifers, cows, and calves in the auction

markets.

Second, the growing importance of auctions in Costa Rica is mainly associated with
farm size and cost of animal transport. In order to have a low (per head) transportation
cost to a slaughterhouse, it is required to have more than 10 head to sell at the same time.
Usually farmers, with the exception of those specialized in finishing operations
("engordadores") sell one or two head at a time at auction markets near their farms. Just
in the Guanacaste region of 10,000 square km, there are seven auction markets. At the
auction markets, buyers of cattle for slaughtering consolidate trucks and ship them to

slaughterhouses.

Third, another aspect of auctions in Costa Rica is that the type of animal sold are
culled cows/bulls for slaughter (not that good quality), usually purchased by small
butchers or intermediaries supplying small butchers, but very seldom there are sold good
quality steers for slaughter. The latter are sold directly to the processing industry and
supermarkets. Also, most cattle finishing operations avoid auctions for selling their
finished cattle (where there are not many buyers for this type of animal). Only during
short periods during extreme scarcity does one see the industry or supermarkets buying at

auctions.

81

Fourth, butcher shops buy mainly from auction markets. These markets work mainly
as a spot market and the main requirements for sellers (producers) is to deliver healthy
animals. Subsequently, butcher shops request the major or the smaller processors to
slaughter the cattle and the shops pay a fee for the service. Note that in Costa Rica the

large-scale processors offer the slaughtering service for a fee to butchers.

(e) Linking Descriptive Analysis with Vertical Coordination Model

We present in Table 9 a summary of different coordination mechanisms strategies
a producer faces when selling to large-scale export-oriented processors, to supermarket
chains or to auction markets. The new relations and partnerships that have emerged
increase the level of control over their coordination mechanisms and reduce coordination
errors in supplies of cattle, ensure better beef quality at competitive costs. We also know
that as one moves from traditional markets (auction markets as a proxy) to large-scale
processors to supermarket chains (with preferred supplier lists) the set of product
standards and transaction requirements become stricter and more direct. Consider first
that the large-scale export-oriented processor supply chain serves a demanding customer,
one based on international (or domestic) safety and taste requirements, and thus
increasing complementary across the supply chain. It also has greater asset specificity as
large plants must have proper supply of cattle or their efficiencies are not
realized. Thus costliness of coordination error is high and well managed/higher-intensity-
of-control coordination strategies are needed. The regional buyer becomes an acceptable
coordination strategy because the buyer understands the processor's needs both in terms
of asset specificity quantity wise and complementarity’s quality wise, and can more
efficiently sort through local supply availability than the processor can on its

own. More intense control strategies would involve significantly more cost (and capital)

82

while not likely yielding much additional reduction in costliness of error. As second-tier
channels, the firms involved likely have less control competence and capital availability
than the leading channel but more than the traditional channel. Thus, some movement
away from spot markets to more managed coordination including a relation-based
alliance. No formal contract exists between the regional buyer and processor transaction,
which is on the border between spot market and specification contract and not a pure spot
market relationship. The buyer will only be successful if he brings the processor cattle
that meet specification needs. In the case of Nicaragua, the processor uses regional
buyers (that previously worked internally but now externally) on a regular basis rather
than transaction by transaction basis. Once a good buyer is found the processor will
continue to use him across many transactions. In the case of Costa Rica, competition for
cattle (because of declining cattle stocks) have become more intense, hence they are

devoting additional resources and avoiding the regional buyer intermediation.

Second, the leading supermarket supply chain probably serves the most
demanding customers (domestically and based on quality sensitivity) and needs to
support its brand reputation/differentiation most intensively-~both of these lead to even
higher levels of complimentary than that needed by the second-tier channel or merely
export-oriented channel. Asset specificity (plant scale and reputation) also raise the
costliness of coordination error. Therefore, this supply chain has the highest costliness of
error. As the leading supply chain market share wise, it is likely controlled by the most
managerially sophisticated and capital-rich firms. The strategy selected can thus be the

most intense from a coordination control perspective. In total then, this supply chain

83

results in the most intense form of coordination in its operation-specification contracts

and an equity alliance.

Third, the auction market supply chain serves the least demanding ultimate
consumer of beef (likely the most price sensitive consumer willing to trade off quality
concerns) therefore complimentary is lowest for this channel across it vertical
levels. Also, being smaller scale at each level likely leads to less asset specificity for
each level. Together then, the costliness of coordination error is least for this supply
chain and thus spot markets are adequate coordination strategies. In addition, these small
players likely have the least capital and control competence (management sophistication
in general) meaning that more intense control strategies to the right on the continuum are
less feasible for them to implement even if some concern about costliness of coordination

error were present.

Lastly, given the differences among the supply chains, our general hypothesis is
that producers’ capacities affect their ability to sell to each of them and that their supply’s
capacities and incentives vary by market channel. Producers selling to the three market
channels are expected to differ in their assets, human capital skills and are expected to

have some impact in their farms’ technology.

3. PRODUCER CHARACTERISTICS BY MARKET CHANNEL

We stratified our farmer sample by supply chains to ensure the sample would
contain producers selling to the supermarket chains. We conducted an in-depth and in-
person structured survey with 300 cattle producers randomly selected from a list of

producers selling to the three market channels in Nicaragua, and to the three market

84

channels in Costa Rica. A complete list of producers was obtained from buyers: (1) in
Nicaragua 180 producers that sell to ICI, to the leading pr'ocessor San Martin, and to the
auction market near Managua; (2) 120 producers in Costa Rica that sell to ICI, to the
leading processor Arreo, and to an auction market in the Guanacaste region.16 From those
lists we randomly selected 10 percent of the producers so the number of sampled units in
each stratum is proportional to the size of the stratum, and that has a sampling error of

10% with a probability of 90%. We collected our data during June and July of 2004.

We start this section by focusing on salient results presented in Table 10. First,
note that the farms which sell directly to supermarkets and large-scale plants tend to be
animal-finishers that have larger farms. Moreover, when comparing the overall farm size
by market channel we observe that producers selling to the supermarket channel are
larger than those selling to the processor channel and traditional channel. Also, the
differences in overall farm size among the market channels are also reflected in both
cattle stock and in cattle sales, hence reflecting the obvious fact that-bigger farms with

more cattle can sell more.

There is a basic reason why larger farmers tend to be on the preferred supplier lists of
the supermarkets and big processors. The small livestock producers do not have sufficient
area to fatten (finish) animals and they sell calves or young steers to the finishers. Note
that in Nicaragua the 96,994 livestock farms that operate in the country have on average
31 Ha and 27 head and in Costa Rica the 30,850 farms have on average 40 ha and 37
head (National Agricultural Census of Nicaragua in 2001 and in Costa Rica in 2000). It is
worth mentioning that if we compare our sample with the national average it is

significantly biased towards the bigger farms. That bias is mainly among the auction

85

farmers which one expects to be nearer the national average size, but are larger because
they are located near and supplying to the most important urban areas rather than the
smaller farms supplying the smaller cities and towns and rural areas. Moreover, note that
in Costa Rica the differences over the strata are bigger because the supermarket
procurement system has been in place for almost 20 years while in Nicaragua it has been

for less than five years.

Second, cattle producers selling to the supermarket channel are more specialized
in beef and finishing production levels as compared to cattle producers selling to
processor and auction market channels. Also, the differences over strata in degree of
specialization are smaller in Nicaragua than Costa Rica. This, combined with the sharper
size differences, suggests that the market changes are inducing differentiation over farms.
On the one hand, those producers that are specialized in milk production have nearly no
presence in the supermarket channel, which makes sense because they only sell calves to
other producers and their tiny presence in sales of animals for slaughtering purposes is
due to their cull cows’ sales, which are lower quality animals not bought by ICI. On the
other hand, as producers that sell to the supermarket channel focus their operations in
delivering the large transaction volumes (that our interviews show are increasing over
time), they need to shift away from cow-calf operations and specialize in finishing

operations given limited resources.

These differences reflect two specific aspects. The producers selling to
supermarket buyers are more specialized in finishing operations and over time we
surmise that their role will be more important as the “market gatekeepers” for the fast

growing supermarket market channels. Those producers not selling to the supermarket

86

channel will become suppliers of calves and young steers ready to enter stocking and

finishing operations, respectively.

Also, producers selling to the supermarket channel specialize more in finishing
operations because cow-calf operations require significant investments in cows and in
land to produce the quantities of calves and steers that are required by the market channel.
This adjustment reflects both limited capital resources and a reallocation of factors to

deliver steers for slaughtering purposes.

Third, as shown in Table l 1, cattle producers selling to processors sell on average
the highest share of their cattle volume to this channel. They are followed by the
producers selling to the supermarket channel. Producers selling to auction markets have
the lowest level of specialization in or reliance on this market. However, note that all
three groups of producers sell to multiple market channels which show the rather
complex market activities of producers. In both Nicaragua and Costa Rica, producers
selling to the processor channel sell on average 80% of their volume to this channel;
producers selling to the supermarket channel sell on average 65% of the volume to
supermarket buyers; producers selling to traditional channels sell on average 47% of the
volume to auction markets. As they can only sell steers to the supermarkets in a
continuous way, in order to sell other animal categories of lower quality (which represent
on average 35% of their sales) they must find alternative markets channels such as
auction markets, intermediaries, butchers and processors. Producers selling to processors
sell a higher share of their volume to processors as this channel will buy all animal
categories, hence it is more flexible in terms of quality and does not necessarily force

producers to find alternative markets. The producers in auction markets are the least

87

dedicated which reflects their “spot market” mode in which they market their cattle

(flexibly adapting to their cattle composition and prices and demands).

Here, for the purposes of classification, we note that producers selling to the
processing market channel also sell to multiple markets but by our definition are thus
classified only if they do not sell to supermarket buyers. Also, producers selling to
auction markets sell to multiple markets but do not sell to either supermarket buyers or

processors.

Fourth, as expected because of its higher intensification of livestock production, the
extraction rate (cattle sales over cattle stock) is higher in Costa Rica as compared to
Nicaragua. The extraction rate, and thus intensification, in Costa Rica is much higher for
the supermarket channel, followed by the processor channel, and third the traditional
channel. In Nicaragua the three market channels show about the same level of extraction
in the three channels, which reflect that as land is cheaper in Nicaragua and
extensification (extensive, land-using systems) is — still - more attractive than

intensification.

Fifth, a higher share of producers selling to the supermarket channel buy more steers
as compared to producers selling to processors and traditional markets in order to fulfill

their buyer’s transaction requirements.

Note in Table 11 that a higher percentage of producers selling to the supermarket
channels buy steers to finish and resell as compared to producers selling to the processor
and traditional channels. As the supermarket channel only buys steers, if producers
selling to this market channel were to depend only on their own calf production, this

would require enormous investments in cows and land even for big farms. As only 50%

88

of the calves born on average are male, in order to guarantee sufficient numbers of
finished animals, the producers would have to buy steers from different sources. Note
that we focus most of the discussion on steer sales as this is the only animal category sold
to the supermarket channel and other animal categories are sold to other market channels.
Producers selling to the supermarket channel are required to increase their cattle supplies

in order to meet the growth in demand of their supermarket buyers.

Also, note that in Nicaragua the most important source of steers purchased (for
finishing operations) by producers selling to the supermarket channels are direct from
other producers. As there is only one auction market located near the capital city,
producers selling to the supermarket channel need to travel on average more than four
times as far as do producers in Costa Rica in order to buy animals from auction markets.
Also, it is possible that producers in Nicaragua are unable to buy good quality young

steers from the auction market.

Sixth, although there is no price difference — for a given type of animal and quality —
over the market channels paid to producers, the main payoff to a given channel is related
to the quality sold to it, and there are distinct differences in quality over the producer
groups in quality. Table 11 shows that the animals sold to supermarkets are of higher
quality. Better quality means younger and heavier animals. In Costa Rica, the genetics of
the animals allow the production of heavier animals and thus larger cuts of meat. In
Nicaragua fattening operations have been very extensive on natural and unimproved
pasture using compensatory growth mechanisms.” This has led to older steers at
slaughter (four years and above) which results in tough meat. Therefore, reducing age at

slaughter means improving quality because the meat is thus tenderer for the consumer. It

89

is not then surprising that in Nicaragua the main focus of producers selling to
supermarket channels has been a reduction of the age (that is, taking fewer years to reach

a certain weight required) of the animal sold for slaughtering.

Lastly, an issue that deserves more attention is the potential link between the weight
quality requirement and animal purchases. Data in both countries support the fact that
producers selling to the supermarket channel buy heavier animals from other producers
for their finishing operations rather than weaned calves steers with 1 year. This finding
indicates that the (heavy-) weight requirements are being passed along the chain to other

producers that operate cow-calf and stocking operations.

4. ECONOMETRIC MODELS TO EXPLAIN MARKET CHANNEL
PARTICIPATION AND STRATA SUPPLY DIFFERENCES

We use a two-stage model in which the first stage is market channel choice and
the second is output supply. Recall that we have discussed at the beginning of the
previous section the method and data used. In the first stage we conceptually frame the
producer’s market channel choice using a random utility model (RUM), which we
consider appropriate for the producer’s decision of whether or not to be part of the
supermarket chains preferred supplier list, or whether to be part of a large processor
supply chain or not. The model is estimated as a multinomial logit with two equations,
one for producers selling to the supermarket chain and one for producers selling to the
large-scale processor. The traditional supply chain is used as the base. After briefly
discussing the main results of the multinomial logit we estimate output supply functions

of the producers selling to each supply chain separately, but control for the producer’s

90

conditional probability of being part of a supply chain by including the inverse mills ratio

(IMR) calculated from the first stage. Then we interpret the results.

(a) Random Utility Model to Determine Producer Channel Choice

We assume that cattle producers select their market channels and supplies based
on their expected utility. The RUM provides a framework that fits the producer’s market
channel selection as it models discrete choice decisions. The RUM resembles an indirect
utility function where an individual with specific characteristics associates an average
utility level each alternative with some characteristics in a choice set. A producer i can
choose among N market channels. Based on our supply chain analysis above, the
producer can be listed in one of the three supply chains, that of the leading supermarket
chain CSU, that of the leading large-scale processor, or the auction market. A producer
can only be part of one of the three supply chains. '8 The producer’s utility from
participating in market channel j is represented by Uij. The producer makes a marginal
benefit-cost calculation based on the utilities achieved by selling to any channel. The

utilities are given by

1 U..= .X..+8.. V'eN
() U fl] 1] U J

We can’t observe the utilities directly but the choice made by the producer reveals
which one provides the greater utility (Greene, 2000). We don’t directly observe the costs
and benefits arising from meeting the standards. The RUM model captures that indirectly
from observing the market choice a producer has selected. Also, as the market channels

have specific standards or range of standards (from qualitative interviews), in practice

91

only a few of them are directly observable by the producer such as: weight, age, gender
but some like fat content are not observable. The survey of farms measured two of them
and shown in table 11 (age and weight). We have not included this in the model because
the channel choice affects clearly influence these variables and endogenous problems
arise. The RUM model reflects the choice made with greatest utility given market
channel alternatives (incorporating indirectly their standards compliance costs and
benefits) given the vector of constraint X (producer, farm characteristics, transaction).

A producer selects market channel k if

(2) Uik >Uij‘v’J ¢k

Here Uij denotes random utility associated with the market channel j, and ,6]. X 1']

be an index function denoting producer i’s average utility associated with this alternative.

The term 81.]. denotes a random error which is specific to a producer’s utility preference

(MCFadden, 1976).19
In our implementation model, market channel choice is modeled with a two

equation system (i=1, 2) as:
3. Y..= .X..+£..
( ) U '6] I] 1]

Where in the first equation Yu=l if producer chooses the CSU-ICI market

channel (Leading supermarket chain), 0 otherwise (which is equivalent to Y1.1 = 1 if

U1.1 > U ijvj ¢ 1). In the second equation Yi2=1 if producer chooses to sell to the large-

92

scale processor market channel (Arreo or San Martin), 0 otherwise. Xij is a vector of

producer characteristics influencing the producer’s indirect utility.

Cattle producer market channel participation is hypothesized to be affected by the
following explanatory variables: AGE, EDUCATION and EXPERIENCE which are the
age (years), education (four categories: no schooling, primary school, high school studies
and university studies) and experience in cattle production (years) of the producer,
respectively. These variables reflect producer’s capacity to understand technologies that
improve cattle quality and transaction requirements and may help a farmer to enter the
supermarket preferred supplier list. Examples are the use of artificial insemination or

summer feeding practices (when prairie supply is low).

HOUSEHSIZE is the number of members living in the household. DEPENDR is
the dependency ratio measured as the number of members in the household with less than
15 years or above 64 years over the total number of members in the household. FEMALE
is equal to one if the producer is a female and zero otherwise. These three variables
combined represent labor structure availability in the household. These variables are
hypothesized to have a positive effect as we expect that the supermarket and processing
market channels use more intense production systems than the traditional channel and

require more labor for activities such as vaccination, and prairie rotation practices.

FARMSIZE is the overall farm size in hectares and it is the sum of land owned,
land obtained for usufruct (can use without paying), rented in, land sharecropped in, less
the sum of land rented out and land sharecropped out. Overall farm size reflects land
availability for cattle production which is closely related to the producer’s capacity to

finish cattle, have bigger cattle stocks and sell more cattle to bigger and fast growing

93

buyers that require more cattle. This variable not only affects the farmer capacity to enter
the supermarket market chain but also its capacity to remain in it. IMPPRAIRI is the
percentage of improved prairies over total prairies. As cattle producers usually replace
natural prairies (e. g. jaragua) by improved prairies (e. g. 'Brizantha, Decumbens, Tanzania)
their supply of pasture increases. This variable is an indicator of prairie quality and also
indicates a producer’s capacities to shorten the time span in which an animal can be ready
for slaughtering purposes. That is, it increases the cattle producer capacity to sell younger
and heavier animals, which are higher valued and also are preferred by the large-scale

processors and supermarket chains with preferred supplier lists.

STOCK is the sum of all cattle categories such as bulls, oxen, heifers, steers, cows,
and female and male calves. FESTOCK is the percentage of female cattle over total cattle
included to control the stock capacity to supply steers. BREEDT is the percentage of
animals that are Boss Indicuss and not crossbred with other Boss Taurus breeds (similar
as in Holloway et al., 2004). Angus, Holstein, and Jersey crosses have higher productivity
levels and higher quality. Although there is no explicit requirement from the buyers, this
variable is expected to influence the producer’s capacity to sell higher quality animals
and be able to select (meet the buyer’s quality requirements) the supermarket channel.
This is because the tropical breeds are less efficient and take longer to reach slaughter age,
in particular when feeding practices are poor. Hence quality may not conform to buyers’
specifications. Note that farmers make adjustments “in the long run” to their cattle level
stocks and to cattle stock characteristics such as how much of it are female or its breed
mix. In the short run (our data) those characteristics are predetermined and so they affect

(or limit) the farmer’s access to market channels that require higher quality.

94

ROAD is the distance to asphalted road measured in kilometers.
ELECTRIFICATION is a dummy variable that represents farm’s access to electrification
which may constrain certain capacities in production such as cutting machines.
EXTENSION is another dummy variable that represents availability of extension services,
which determines the producers’ access to production techniques such as sanitary
services and marketing options. These three variables capture farm’s access to
infrastructure and the larger the distance to roads (it is hypothesized) the less likely is a
cattle producer to sell to supermarket channels reflecting difficulties in trading important
quantities of cattle consistently. The availability of ELECTRICITY and technical
assistance are hypothesized to affect positively the odds that the producer can implement
correctly some technologies that will produce cattle with quality levels desired in order to
sell to the supermarket channel. COSTARICA is a dummy equal to one if the cattle
producer is in Costa Rica and zero if located in Nicaragua and it expected to capture

differences in the industry and retail sector between the two countries.

Table 12 contains the market channel choice estimation results. Overall farm
size is a highly significant determinant of market channel participation in the
supermarket and in the processor market channels. We know that larger farms are more
likely to sell to these channels than are smaller ones. Although we would need
additional data before concluding (complete data for several years, which unfortunately
unavailable), this might indicate that as the farms increase their operations they are
more likely to sell to these two market channels. The magnitude of overall farm size is
about ten times larger for producers selling to the supermarket channel as compared to

producers selling to the processor channel. This suggests ihat the supermarket preferred

95

supplier list favors bigger farms that can deliver more quantities of cattle or that farms
selling to supermarkets have better growth prospects. Although in the short term,
overall farm size can be considered fixed, in the long run endogeneity problems can
arise from the fact that farms that participate in the supermarket channel can grow
larger. The later is more evident from the qualitative analysis data that supports that
superrnarket’s increasing cattle quantity requirements are not to be met by exclusively
adding new farmers but rather as a combination of new farmers in their preferred lists

and more important “higher quantities” from the current producers in their lists.

Also, if a producer is located in Costa Rica, his/her odds to sell to the
supermarket channel and processor market channels are lower as compared to a
producer located in Nicaragua. This is because there are fewer producers in the
preferred supplier system of the leading supermarket chain and of the major processor
in Costa Rica. Also, the development of the auction markets in Costa Rica indicates
that producers have greater access to other market channels. Lastly, the development of
the supermarket sector and processing industry indicate that their quality standards are
stricter and it is harder in Costa Rica to sell to these market channels as compared to
Nicaragua, where these market channels are just initiating the introduction of private

quality standards.

The asset specificity of prairies and cattle stock breed does not affect
significantly the probabilities of producers’ selling to the supermarket and processing
channel. One would think that these variables would affect positively their capacity to
deliver the supplies, and the quality required by the buyer’s technical specifications

(minimum weight and maximum age), hence favor the access to the supermarket

96

preferred supplier list. Perhaps improved prairies and cattle stock breed variables will
become a significant factor in the future once marginal quality gains from other factors

are fully exploited (such as carcass weight and age).

Fourth, human capital and labor availability do not have significant effects. The
only two variables significant for the processor channel are education and dependency
ratio. Results indicate that more education reduces the probabilities of selling to the
processor market channel. Note that the negative correlation between education and the
processing channel was unexpected. See that there are greater differences in producer’s
education among the channels in Costa Rica as compared to those in Nicaragua. As in
the case of Costa Rica where we observe greater difference in producer’s average
education levels among channels, the processing channels has the lowest level and
those selling to the supermarket and auction channels have greater. The processing
channel has been operating since the 603 and 703 and the auction and supermarket
sector started in the mid 803 and 903, which perhaps explains greater differences in
educations between the channels in the two countries. It is possible that those more
educated are able to understand and compare the requirements arising from new and
multiple market channels and capture the benefits of selling to them. This may indicate
that those more educated are shifting away from selling to the consolidated processing
sector and moving towards alternative market channels. Higher dependency in the
household increases the probabilities of selling to the processor market channel. Lastly,
the only infrastructure variable that affects positively the probabilities of selling to the

processor market channel is electrification.

97

(b) Market Channel Output Supply functions

As mentioned, we use a two-stage approach to specify our model. In the first
stage the model uses a random utility model to frame the market channel decision choice
and here we present the second stage of quantities supplied. Note first that some of the
previous work exploring the set of factors affecting the decisions of market participation
and the quantities traded include Goetz (1992) in coarse grain markets in the southeastern
region of Senegal, and Holloway et al. (2004) for milk producers in Ethiopian markets.
Using data from Kenya and Ethiopia on livestock markets, Bellemare and Barrett (2004)
model the two decisions comparing sequential and simultaneous decisions and find that
these decisions are made sequentially. In our work, we estimated such decisions
(participation, and quantity supplied) as both simultaneous and sequential and did not
find major differences between the model results. We do however, prefer the two stage
approach because it helps to control the conditional probability of a producer being in a
channel and it potentially control unobservable characteristics that might generate self-
selection problems, hence we present the model and report results of the two stage

(sequential) estimation.20

In this section we estimate the output supply function of cattle (or marketing
supply function). We derive the equations from the profit function using Shephards

Lemma and obtain the following two marketing supply functions:

(4) 6.+6.P +6.W +6.K..+6.A..+n
J m J m J 'J J 1]

Yij(males) = j ij

=a.+a.P +a.W +a.K..+a.A..+7]
J m J U J 'J

(5) Yij(female) J J m if

98

Note that before, Yij = (0,1) represented market channels and now, Yij = Number
of cattle.

Where the ith producer sells to the jth market channel represented as mentioned
before by the leading supermarket chain, large-scale processor, and auction market
supply chains. YiJ-(males) are male cattle sales and Yij (female) are female cattle sales measured
in number of heads sold for slaughtering purposes. Note that we disaggregated the animal
categories following previous work that provides insights on how the supplies may
change depending on how the supply model is specified. For example, Jarvis’s (1974)
seminal work stressed the need for an econometric model to disaggregate by animal
categories because cattle are considered consumption or capital goods depending on their
age and sex. The disaggregating is further discussed in Aadland and Bailey (2001) where
the ranchers problem is specified taking into account if the animal category is fed or
unfed.

P is the average price of cattle measured as USS per head. 2" It includes two prices,
STEERP is the average price of male cattle sales and COWP is the average price of
female cattle sales in the municipality where producers are located. W represents the
vector of input prices, however it only includes the variable LANDP, which is the
average land price of buying one Ha (U S$/Ha/municipality). Other input prices were not
possible to include because of lack of variation or because producers tend to use a high
variety of inputs, which we did not have prices for. K is the vector of fixed factors and
shifters that include the following previously defined (in the first stage) variables: AGE,
EDUCATION, EXPERIENCE, HOUSEHSIZE, DEPENDR, FEMALE, FARMSIZE,

STOCK, ROAD, ELECTRIFICATION, EXTENSION and COSTARICA. 41.942. are the

99

9

inverse mills ratio (IMR) calculated from the first stage used to control for the producers

conditional probability of being in a channel.22 6]. ,aj. are the estimation parameters and
771.]. is the error term. We estimate the system for each market channel using a Zellner’s

seemingly unrelated regression (SUR) model to exploit potential correlation across the

errors in the three equations in order to gain efficiency in the estimation.

Table 13 presents the results for the marketing equations estimations.
Elasticities are rather small as expected in the short run and are in line with economic
theory (as oppose to long run where fixed factors can be adjusted and elasticities
increase) and previous aggregate model results, which has shown even negative
elasticities in the short run (Aadland and Bailey, 2001). There are four significant

results that appear most relevant.

First, supply elasticity of males with respect to the cattle stock of producers
selling to the three market channels are similar and equal to 0.08 and for female cattle
supplies it is nearly zero. This finding is not unusual because cattle stock decisions
involve long term periods as shown by the rather abundant literature on cattle cycles,
which indicate that cattle stock varies smoothly over times showing peaks intervals of
approximately five years. The cattle cycles in part have been linked to biological
production functions. The model suggest that supplies marketed might not increase
substantially as cattle stocks and overall farm size increase, which reflect the limited
adjustments farms can make to their cattle stock and overall farm size in the short run.
As expected female cattle stock has a much less important impact effect on supplies

given the time period it requires to produce calves rather than liquidating productive

100

stock. Producers may not sell their current cattle stocks to. increase their current
supplies, and as expected this is more important for female cattle considered capital
goods. Also, the estimated parameters, suggest that producers selling to the three supply
chains do not differ much in their capacities and technologies to increase their supplies
with increases in their cattle stock. That, however, does not mean that cattle stock levels
are rather important to increase their supplies. On the other hand, producers may use
other strategies to increase their supplies such as buying more animals from other
producers rather than slaughtering their current stock. As suggested by the descriptive
analysis, in relative terms more producers selling to the supermarket chains do buy
animals to finish as compared to producers in the other supply chains and they are more
specialized in finishing operations. The inelastic supplies of male and female cattle
supplies with respect to cattle stock, possibly helps to explain why producers in the
supermarket chain are more specialized in finishing operations and buying more cattle
from other producers as compared to producers in the large-scale processors and
auctions markets supply chains. More research will be required with times series data in

order to validate these results.

Second, results show that producers selling to the supermarket channel have a
price supply elasticity of male sales that is four times bigger than the price elasticity of
producers selling to the processor channel and 12 times bigger than producers selling to
auction markets. We expected indirectly this result because producers selling to the
supermarket chain are required to meet rapidly the quantities demanded, which is a
requirement (keep up with demand) for being in the preferred supplier list. Now, there

are a few things to consider, we may not safely claim that producers respond

101

exclusively to the price increases. The price increases could mainly reflect seasonal
price increases, but the producers selling to the supermarket chains are literally forced
(or risk themselves to be de-listed) to supply the required quantities of cattle. This is
indeed the main advantage of the more integrated supply chain developed by the
supermarket chains since they can ensure their cattle supplies in times of scarcity and
with the plus of getting the higher quality animals. This is of course most important for
male cattle because the chains do not buy the female cattle, considered of lower quality
(cull cows). We think that producers selling to the supermarket chains have specialized
in finishing operations and stimulate other markets such as buying from other producers
in Nicaragua and from auction markets in Costa Rica, in order to respond to the chains
minimal supplies requirement (transaction), which is rather most important in times of
supply scarcity. The chains ensure a smooth supply of cattle meeting their stricter
quality requirements while the producers in their preferred supplier list benefit from
selling continuously over the year (and increasing their market over time) but they have
to adjust and innovate in their strategies to stay in the supply chain. There seems to be
indeed important potential positive effects over cattle markets, since the finishing
operations may become important buyers of cattle from other producers or from auction
markets, where the latter have skyrocketed in Costa Rica at the same time supermarkets

skyrocketed.

Third, the supply elasticity of female sales with respect to the female cattle price
is only significant and negative for producers selling to the processor channel. Here we

make a contribution to support the negative short run price elasticity of female cattle

102

(capital goods) and position ourselves in the already long debate of whether short run

price supply elasticity are positive or negative.

Fourth, there is a marginal absolute increase in the supplies of steers as a result
of an increase in the cattle stock, overall farm size, and the price of steers for cattle
producers selling to the supermarket channel. That is on average two times bigger than
the marginal absolute increase of cattle producers selling to the processor and auction
markets. This is a reflection of course, of bigger farms entering the preferred supplier
list of the chains as compared to producers in the processing and auction market supply

chains.

5. CONCLUSIONS AND IMPLICATIONS

We first present the conclusions and implications that arise from the quantitative
analysis and second from the supply chains qualitative analysis. Note that we used cross
sectional data, hence our results should be taken with some caution, in particularly in

1
terms of supermarket causality effects. They are useful to generate hypothesis to be tested

and validated by additional research.

Implications from producer ’3 market channel determinants and supplies.
Results show that when producers sell to the supermarket channel they face similar
determinants as when selling to major export-oriented processors. Overall farm size is an
important determinant and raises questions as what will happen as the supermarket sector
grows and gain participation in the domestic markets. Second, the supply elasticity and
absolute marginal response of producers selling to the supermarket channel are higher as

compared to producers selling to processor and auction market channels. This potentially

103

is a result of the buyer year around supply requirements. Third, producers in the
supermarket supply chains tend to be more specialized in 'finishing operations which
seems to be a reflection of the adjustment that they need in order to remain in the chains
preferred supplier list. Subsequently they will become important buyers of cattle from
other producers. Although we acknowledge that more research is needed to verify our
findings, they seem to indicate that bigger, specialized and more responsive cattle
producers are likely to become the market gatekeepers in the arising preferred supplier

list of the more integrate supermarket supply chains.

Smaller producers, supermarket chains and auction markets. From the
interviews we found that the supermarket chains’ procurement systems shifted away from
auction markets to avoid the additional brokering cost (5%) and avoid sub-optimal
sanitary conditions and animals handling (stress). The later shift suggests that as the
supermarket sector grows and displace the traditional retailers, auction markets will see
their demands shrinking overtime. However, auction markets in Costa Rica have
increased substantially in the same period the supermarket sector grew. Our data shows
that more producers selling to the supermarket chain “buy cattle” (as oppose to raise their
own cattle) and their main source are the auction markets. That said, as supermarket grow
auction markets might become important sellers of cattle to producers that sell to
supermarket chains rather than selling directly to the chains. Our data shows that smaller
producers sell to auction markets as compared to supermarket chains and processors. Our
results in Nicaragua indicate that the producers selling to the supermarket chain use as the
main source of cattle other producers, which may indicate just the absence or incipient

development of auction markets. From the Costa Rican experience, we might not

104

necessarily conclude that as supermarket develop in Nicaragua auction markets
development might be impeded (because the chains do not source from them). What
happens to the auction markets is an important question because our data supports that
smaller producers sell to them. Moreover, several other. authors have indicated the
positive effect of developing auction markets in Nicaragua, which possibly may replace
the intense trading activity between producers in Nicaragua. Auction markets may help to
reduce the transaction costs of trade between producers, also provide higher transparency
in price determination, and may help the traditional butchers in Nicaragua as is the case

in Costa Rica.

Effects on the beef processing industry. As supermarkets buy only cattle carcasses or
process their animals in large-scale HACCP certified processors and do not buy from the
smaller processors with insufficient food safety processing capabilities (both in Costa
Rica and Nicaragua), the effects on the beef processing industry indirectly accelerate the
disappearance of smaller processors. Even when we don’t have data that characterizes the
producers selling to the smaller processors; it is somewhat obvious that the low volumes
processed by hundreds of small processors mostly in rural areas are sourced from smaller
producers, which is yet to test and validate by additional research. Moreover,
supermarket chains’ investment in distribution centers and HACCP processing plants for
further processing are increasing their capacities as compared to traditional butchers.
Also, supermarket chains have become competitors for cattle which affect major
processor procurement systems and also affect auction markets. The general implication

is that consumers are benefiting from higher hygienic and safety standards by shopping at

105

b

the supermarkets but the smaller processors and producers that mainly sell to them will

face a declining/shrinking demand for their cattle.

Changes in the beef supply chains coordination mechanisms. Some supermarket
chains rely on a dedicated wholesaler to handle all cattle/beef operations or may have an
internal department that undertakes this role. The supermarket chains have implemented
higher quality standards when buying from cattle producers in order to improve the
quality of their beef cuts. Also, the leading chains have developed a preferred supplier
system in order to buy from cattle producers rather than from the major processors and
only contract with the processor to slaughter the cattle.

The change in coordination strategy offers the advantages of getting enough supplies
of cattle during seasonal scarcity periods and with the right quality. Producers benefit
from an increasing year-around market that offers competitive prices. On one hand, the
more integrate supermarket supply chain benefits the consumers as it enforces its quality
standards, thus ensuring its reputation in the mind of consumers. On the other hand,
higher levels of control over the supply chains are likely to set stricter rules in which

those producers that don’t live up to the standards will be‘excluded.

106

2300: one mafia": mo 3mg 0:1? 05 m: :55 was xenon _: 28% me 8:023: g :0 :80 2: ma :23.on

m: 8:: m: ”macaw mwgmmb m: 8:9: 03:9; cameos: a. oEmeO ”9:: Hone—\mmbv an: .8: own? : “one: ”—05:30 3 omega
2:83 0:: m: 8:: 58mm Gav :5 3:9: .3 32:: 8:: 8:88 38: c2838: 92:9 .0: A>m<v 0:13 6:83 omega 3:9:
A0 ”out: 9an: omega 295m 520 wave—:2: 05 88: 83: 3:38:93: 3053: 3:32: owfio>m 3:9: :20 ware—:2:
2.: 2::3 Crag/U 0:3.» 8:on 359:8 ESE: ANV mow—20>: 2:58 o:_a> 59: £20 maze—.88 05 .3 32:: Acouoicoa
92:05 59:: omega 2:83 :30 wane—:2: 2: 3:5 8:833 5688:: 5:383: m:_w:oQ-::oo 8:25.: 5538:
E30 9:8me 3:9: 0:: 9 3:3 m: $36 533038: 3:93: 12:32: omega Ego mazes—:2: 3:9: 05. 2v ”8an

 

107

 

”a... 2% £3 83 3.0 ”2.: a} 3.2 2% a: 23
~ _ a o _ an m. a S I . am: 5 :88 was:
1 R. _ E 2 c M: am 2 3% 822:3:
2 § 5 M: E m 2 8». :. Ev: 82033:
mm mm m 2 mm a K S 2 3% 822803
w mm m- 0 mm m 2 a : av: :28
3:9: 03:33 59:75: £532:
2 8 _ no mm 3 2 cm 0 3% 28:5:
3:9: 23E? 8:2 no: oEmmHO
mm mm 2 2 mm on 3 S S 3% 5%: BE a. base :83
3th 03.5 «3.5
.32"..— >._< Z: .33... >5, Z: .38..— >._< 2: «3:22, 2:...—
.2553; 3.3.5 3 .3553:ch

 

$895: 8:83:89 mflmgm foo owfim
988m :8: @3233“: EEO wave—:22 .3 A>m<v 33> 8:605 $825. 98 A>A=>U 0:136:85 3:932 K. 2:3.

N 033‘ 03 :25293: .32 :8: 8:35am? 8.: 88:2

 

 

:3” 83 $3 :5 83 2.3 $3 a? 2% 2.55%
ox...“ wfi.“ 2.82: .25.: :25: E:— 0885 .02
3a was 33 ”a: $3 $3 $3 2% 82m «2% £3 E 2.52: 2:8... :z
o: 3.3 83 _8.~ :3 $3 23: . $3 8.3 $3 23. 2252: .80 .35
a _ E 23 2:. ”a. ”8 8m ”2 8m 2% 3.53: 38 3%. 8::
05.“ Ba x: a: N: R 3 2:. an 2: ”R 2:52: 38 .8: 2:3:
as _ 2: 8e 8: 2K 2:. 2: N? E. m _ w a: 2.52: :8 .33
56 2x:
0 .3? 3: a a 8 8 8 2 :3 E 5:325 5? 22.9% 83 :0 22a
oo oo— 2: 2x: ESP?
3 2: 2: Na 2: 2: Saar: a? 2832: :o 2:52
cosmwt:
N.
a 3:: _ - m - co - - - 2x: .838: 3:52: 33
.3. x S om
22% S 3 S 8 - Q 22: 8:8: .955 a:
a 55 .m ow cc.
as .23 E m on R a 22.: .98: a:
2 - 8
n .3 _ - m a - : 22.: 3:252» 3:
co: 0::
a: a: .35 002-52 0:: .55 -82 0oz =2; -82 ouz
852:2? .2523; 2.3.5 3 3:25:58

 

28:33:80: 35:88:30 8:83:05 03:89 6 03:;

108

 

 

 

n: 2: a. a: o: E 3; 5 9: :53 ems :5 :2: E.
a NN om mm mm mm mm mm mm 5 22:83 :8 “5:;
E N: m: ov a. $ “8 3; MS Ammama 22% £830 :8
AaEmmB
3 mm .3 :m :m 5 SN 2N 8N 22.an we; 2;. 8x230 .80
:~ ;~ .3 N? as. 8m 5 SN 3: 35328229: mo :8
mm 3. R S No a. 2 : 3 333332055: ho moo
fim 8m «2 £0 E SN :5 :m 80 azamamoEBEmE E :6
E SN as o: w: or 3. e: m: 2:53 2&8 go :8
SN 8m EN EN 3” :N N2 8. Rm @332 waif? 3.8
A3 mania—.35
and odd 8} an; 23 ~34 35 E: :3 ES mg 335 932:; 3%.-..2
an.“ 2%... ME 3 8m 2% 3:. S m as a N: 5 €583 :8 5&5;
N e
a,“ ”a wad Roam SEN 3.2 812 a: 83m 22m 833. 953: 22>
_
0 $3 25 5. m3 So one 26 m3 2 8o 5 35me sneeze SE
ESQ:
an 0 SN.“ 3% 23 8o.» 39¢ Sew $3 :3 33 32.83.. 8.3.385 2.5
on 3. on 8 a m S h on Essa 9.3.6.5. 3 his:
002 002
5 as .35 0025: RE .55 ._.oz ouz .85 .82 002
92.3555 .2223; 3.3.5 3 2:55:58

 

mosmtfioflmso 8:26on Snack “v 28;

109

.0 0.000

00m 8.. 000 a. .00. 008.. 000080 8088 .0 00:0 .008 000 00>0 :0..0w.:. 5.3 8088 .0 00.0 .008 0. 5.008.... 00.3 8088 .0 00:0 .008 .0 080m 3. 20.. :00.
8000 .000. .008 0:0 80.00 80:. 0000.80: .8000 w:..... .8000 0:000:08 .0 5.8883. 000 000. 0800:. 808 00. .0. .000 8.0:. 080.800. 00. .0080. 08 0. 0800:.
002 A... 8:08.008 .:0.000_..000.0 8800030.. 88:88 ..0.::00 0003 80.0088... 80.0%.... .:0.000...:0. 8880.0 .w:.:00.0 0:0. 00 0000 00.0.3000 :0000080

:. 0:000 0200 0000-. A800. 00:80. 8 .000 0.5 8.808.. 0w03 00.... 008.0 0.000 .000. 3.80. 0:0 00.... :0 .0 5.88800 05 0. 8000 800-. G. 3:088:00

08 80000.0 8 000: 0000. :0. 0.00. 0000 8:00.000 00:0 0:... 0.. :0 ...0 8 80808.00 w:..... 0:... .0880 80:0 .800: 0 8.80. .0 00.00. 000 0. 0000 w:....-.- .3

.00 .00. 000000. 8050 0:0 0.0.0808 w:..88 .000.0.w:=. 0020.000: 0006.608. 0.0.3. 800.....0. .0 000.000.000.800: :0 .0 5.008800 08 0. 80:. 0.00..0> .000.
-:02 .3 3080 0... 8 008088 05 w:.._0>..00 :. 00800:. 00 .00. 8000 0:000:08 =0 .0 5.008800 08 0. 0000 0:000:00... Am. x0533. 000.080 0:0. 8088 .88
>0 000.>.0 8:00.000 :0 800.008 :0 00.. 00.00. 008.. 0.00 8088 .0 b.8000 0. 0800:. 0005 .08:.0> .00 08.3.03 0:00000 :0 800.008 :0 .0. 00.8 0w000>0 000 0.
00... AN. .00. 008.0 8000 8.000.008 00... :. 0.000080. :0 .0 5.000080. .08 05 00>0 0002 .00 .0: :0 080.000 005.0 8000 8.000.008 :0>.w 0 :. 8000080. :0 .0
5.80080. 8088 .008 0... .00 0080008 00000 000 0. 5.000008 8088 .00 .0::000-0:0 .0 080m .00. 008.0 .0::000 8.000.008 .00. :. 0.000080. .0 0008:: .008
000 00>0 0002 .00 080.000 00: .0 080.000 8000 8.000.008 :0>.w 0 :. 0.000080. .0 0008:: .008 08 0. .x. 0.000008 .0 0008:: .00 .0::000-0:0 .0 080m 2. "0807.

 

.0000 000030.. 0.00: 00000-80:—
.000 0000.000. 0003

0050.000. .000. 00....
0000000 .000. 00.000
0.00000. 0000-.
A05 8088 .0 008 .80..-
5.8000... 8088 .0 00.0.00

.00.
.800. :2. 0920.00. 00.08—00.00. 8080..

 

00.0 00.0
0.00 0.0.0 00.0 00.0 00.0 00.0 00.0 00.0 0.0
00.0 0.0 00.0 00.0 00.0 00.0 00.. 00.. 00..
.00 000
_ 000 000 000 000 000 000 000
00 00.
0.0.0 00.0 00 00 00 .0 0. 00 000
.00 0:.
0.00 0.0.0.0 0.0 000 000 000 000 .00 000
.. 0.. 0.0
0.0.0 0.00.00 0.. 0.. 0.. 0.0 0.0 0.. 0.0
_. 0.. 0.0
0.0.0 0.0.00.0 0.. 0.. 0.. 0.0 .0 0.. 0.0
0 0
0.0.0 00.0.0.0 0:00 00.00 000.00 000.00. 000.00. 00.00 00. 0.0.00 000.0.
002 002 002
.0. .0. .000.- -._02 002 .0000 -002 802 .0000 -002 802
80008005 .8202; 0.00.00 3 00:80:08

 

0:0..0::.0:0U% 00.00.00.00800 5.000080. 8080.. ”m 0.00.-

110

 

om mm 2 mm 2 R 3 2 2 9:35 mosawcfi
m m m n n m N _ m 3:qu 32°33:
: S o. 2 2 2 z w 2 3:3: 822335
8 ”N 2 on «N a 2 : 2 3:3: ”56m
8» ma v? 25 ”8 Sa 32 CE 32 azuxv 305:5“
3: E
San—-5:— 3 3:53.. .37.»...— owflo><
: : a 2 2 : 3 om 2 $233 0222.: mo 85
2 2 o 2 2 2 2 2 2 vacmmasaoss: .6 8:;
mm 2 «m R 2. 2 3 on 3 Amémmamosesas 2o 8E
w b a o e v 2 o. E 352: ago“. he BE
one 25 3o ”3 one NS «No 25 ”do 3:33 202:5 2o 85
Gimme
3:9: Lona—-5: ha 3:.— $9.22
2. _.w.§ 83 53 :3 ~22 an} 9.3 $3 23 82 A252: .89 .38.
2 3 3 on a m 2. u 2. @553 e839... 3 .352
ouz RE 002
5 a .35 -Sz 00: .55 .22 0oz 3.; .52 RE
3.55:3 .2862... 2.3.5 3 3:38:58

 

mormtfiofimso :28:on 8“th ”m 2an

111

02.22080- Q 2:200 02 :00500 n 0 0:0 2002200:- Q _0=c_t0:-Dm0 :003000 N 0 2:200 02 Q 0:2:02-0m0 :00500 n 0 H220: 02000
3:00.02: 303:3 00:85:20 2.0— :0 0000-0. AC 002-:02 02.220000. Q 002-:02 2:200 02 :00500 H _ 0:0 ”002-:02 3:22020- Q 002-:02
005:02-0m0, :00500 n 0 002-:02 2:200 02 Q 002-:02 00.50.02-Dm0 :00500 n w .002 3:220:00- Q 002 2:200 02 :00500 "0 .002

3:220:20- Q 002 .0P0Eom-Dm0 :003000 N 0 .002 2:200 02 Q 002 _0:b_tom-Dm0 :00300 n 0 002-:02 3:220:20- Q 002 2.2200;

:00500 n 0 002-:02 2:200 02 Q 002 2:200 0.0 :00500 n 0 002-:02 0:5:02-3m0 Q 002 0:2:02-3m0 :003000 n 0 U20::000

-000 002-:02 m> 002 0003000 00:02-02:20 $0. 00 0000-0- 30 0:52 05 022:0 E0E>20E0 023 0:00:08 00 0008:: 05 000 00080:. >5 8 002-000
:000 22.. 06002 E30 000 000 0.000000: $0 ”.2: 00E: 020 20000000 _200 00>0 200» we :05 000.0 :0 200» 2 :20 $2 023 0.0000000 000 2 20080:.
.«0 00:3: 000 2 2:2 5000:0000 $0 0200» 0.0:. 00 m. 023 200508 0.0000000 00 0005:: 0.0 2 2_=0< $0 220:2: 0 022 05 w250v 20002.00
2.000000 05 2 m2>= E00808 £500 00 00:5: :29 000 2 02m 0.000200: ANV ASSN cow-000 @0020 .0000 0:0 000000000000 0:2 0:0 0:0 000:0:

0:2 .00 :000883 000 $2 2 00000000000 200 2 000:0: 220 002002. :00 002200 0:2 2:0 00:30 0:2 00 :22000 000 2 020 5:00 =S0>0 A: 3.0.02

 

 

 

a.“ $0.0 2. on S 2. mm 3 3 S 2: 2.08.52303:538:80:
E.

n. 3.00 we 8 3 5 a 8 3 z 3 22020203380:

2 3 2 00 2 2: 00 02 S 2.0 20:00.0 8:

man—swam.

0.3.. 0,0,3 m m 0 2 2 m 2 _ : 050.008.02.208 98.500
07.0

Eu 5:5 002 002 002

3.: .ea 5.; .82 002 .35 .82 002 3.; .82 ouz

3:32:20 .2223; 2.3.5 3 02:20:00:

 

A:000::0:00v 0000200000000 8.80 ”m 200,—-

112

Table 12: Producers' market channel determinants

 

 

Number of obs = 300
Wald chi2 (30) = 228.55 Prob > chi2 = 0.0000
Log pseudo-likelihood = -l46.19968 Pseudo R2 = 0.4361
Covariate Supermarket Processor
Age of farmer (0.01) (0.01)
(0.02) (0.02)
Female producer -0.61 -1.09
(1.23) (0.79)
Producer education 033 -0.41 1'”
(0.23) (0.19)
Experience in cattle production 0.00 0.01
(0.02) (0.02)
Household size -0.08 0.02
(0.12) (0.11)
Dependency ratio 0.20 1.342"
(0.70) (0.61)
Overall farm size 0049*." 0005*"
(0.010) (0.049)
% of improved prairies -O.734 -0.004
(0.65) (0.55)
Cattle stock 0.0010 0.0001
(0.009) (0.001)
% of Indigenous breed only 0.004 0.005
(0.007) (0.006)
% of female cattle stock -0.011 0.002
(0.009) (0.008)
Electrification 0.332 1.158*
(0.736) (0.665)
Distance to asphalted road (Km). 00] 0.01
(0.02) (0.02)
Extension service «0.26 ~0.48
(0.51) (0.42)
Costa Rica -1 .909*** -2.126***
(0.73) (0.70)

 

Standard errors in parentheses

* Significant at 10%, ** significant at 5%, *** significant at 1%.

113

 

 

0.20 0. N 0 m 0 . .. 0 .. 2.8... 0.0000 0.000
0.0.0 00 2 S R 0m 0m 8 .0 E .20 5005...... 0:00... 00..
0.0 0.0.0 2.0 00.0 2.0 2.0 N00 000 2.0 00.0 3 .0 00.00... :5. .00 .23 20050:. 0.2.0000:
0 0.0.0 2 MN 00 2. 0m 0m 2 0 .. .000 0.08... 0:0. .00 00.. 0.000000:
0.0 ...0.0 P 00 00 R 00 0.. 00 00. 3 .20 .0030... 0000000000 .23 0000. 00.00:
0 0m mm 00 NM 2 .0 0m mm 00 .000 0.00. 2000000000
0.0.0 : 0 mm .. m 0 0 ... 0 .20 .8000... 0.0.00.
0.20 m 0 m m 0 v n m: 0 32:00:00.. 00 0.003 5000.00 00:00.0
.. . .. .. -
0.0 0.0.0.0 2 2 0. ... 2 2 0 N 0 .2000 000.089 .23 00000000....
0.0 2.0... 2. 3. 0.. 00 R 3. 5. an 2. 0.000 .8000... .0 02.
0.0 .00 2.. . .3. 00.0 mm... 00.. 0...: a... E: 2.00.200.m .50. 00.300 0.00. 0003-000:
0.0 0.0 .0. Z 0.... 0.0 0:. 0.0 .0 0.0 0.0.0.0...
0
00.0.0 S. 0:. S. 0:. .:. 0.0 S. E .0 0.000.020 00:0 0.2.020...
0.. n: N: N: m: 0.0 m: ...0 m: .05 0.0.0.000 .000. 00....
:0 0.0 0.0 .: 0.. 0.. 0.0 0.0 0.0 .0... 0055000., .200 .0 .30. 00....
..: 0: 0. . 2 ... m 0: 0.0 m: 0.0 .0... 0.2.9.0 .50. 002
2 N... 2 0.0 0.0 0... mm .0 2 .0... .000 0. 00000.0 00... .0000
0.0 0.0 0:. 0.2 .0. m... 0.0 0.0 0.0 0.0 :0 .05 00.... 8.0.. .205
.20
00. 2 0m 00. 0.. . . 00. 2. 00 0000000... 0.050. 2. .05293000 20:0
000 2000020
00. mm 2. 00. .0 0. 00: 0. 0.. .0 .0050: 2.. .00000900000 0.000
00 0.. 0m 0m .0 0 0.. 0 0m .0 000.500 2000.00... .0 .0050:
.00 0.0
E 0 00.2.0 002 002 002
.35 .000 .500 -002 007. .300 -00: 00: .50... -002 002
000000.020 .2203... 0.00.00 0.. .00...:0:-000

 

0000200002000 8.00 ”N 2000.

114

 

0.52.800: 0% 0200606085 cooéom Am
02.860808:—

£ 08.00:. 2000—0:3 33:0— :ooEom AN

08—00:. 20022.3

23:0— 0.... 03:82 10:20:00. H Season 2
.526

0.0000 0.0.00. 0.0.30: .2203; 2F
008385

00. 0020022.? 30:06on 03,—. :oozcom AN
000—0020;...» 30:28am 95,—. 0% Ease
aoxaEhoazm 9.808 0.0200 n..— cooEom 2
"52.0 ban-5 3.058 €25.00

0‘: 0.52.0 «Sta-Evaam ..o_.—..._.=ouom
0.8.6200 0%

£20022...» cod—063$ E0505: :oozzom AN
€200,233 0320.038 3:595:

00. 3.23001050250 $5000: DmU 523.5 C
"5.2.0 5.5.5

32:3 5mm: 0.52.0 «Stanton—i mica?—

 

.3335...—
.3330»

3.5:: 025:? 30.5.50 03.0.02

68.3 30.3 0:3 3am
.. 33.0”.— - .3323— 0.359on

 

$825 =23£Ecou .0323»

0:38.005... 52.0 .0395

 

050:0 bamzm 80th 82E. 2: E moWofibm :otmamEooU .momto> “_ 030%

115

CHAPTER 4: CONCLUSIONS

The development of supermarket procurement systems mean for producers
market channels that have implemented more integrated coordination mechanisms and
institutional arrangements with higher levels of control over the supply chains. Some
supermarket chains rely on a dedicated/specialized wholesaler to handle either all
produce or all cattle/beef operations. In the more integrated systems, enforcement of
minimum levels of quality attributes and food safety standards are implemented and
consistency of supplies are both coordinated and transposed to wholesalers and producers.
Investments such as distribution and logistical systems are shifted upstream. As
producers sell to supermarkets, they become part of a preferred supplier system where
producers enter and coordinate a continuous supply of “the group” as oppose to their

independent individual supply to spot markets.

Changes in the retail markets as reflected by the increase in market share of
supermarkets do not necessarily mean that alternatives markets may not develop. In
Costa Rica, and specific to the cattle supply chains, even when the supermarket sector
increased substantially its share in overall food retail and the supermarket chains
procurement systems shifted away from cattle auction markets (to avoid the additional

brokering cost animal’s stress), auction markets skyrocketed during the same period.

Although there are substantial requirements and significant payoffs to entering the
supermarket channel as compared to the traditional market channel, it varies by product.
In the case of cattle, scale and size do matter and there are only few producers in the

supermarkets’ preferred supplier list. This raises the question of what is going to happen

116

to the several hundreds of producers selling to traditional markets as supermarkets gain
market share. Results show that when producers sell to the supermarket channel they face

similar determinants as when selling to major export-oriented processors.

In the case of tomatoes, growers that sell to supermarket market channels face
higher costs, stricter conditions and control, slightly higher prices, and technology with
higher yields. The growers selling to supermarkets generate higher net incomes per
hectare as compared to traditional market channels. There is no conclusive finding in
terms of their technology type as while the leading chain is labor bias, the second tier is

capital bias as compared to traditional channel.

Small farmers can access the supermarket channel if public/private support is
available to them as well as organizational capital. However, there are substantial
subsidies in the form of technical assistance, soft credit access, access to fixed assets in
production and marketing. The challenge is to design or provide technical assistance in a
way that the “producers access become sustainable overtime”. Even when the volume of
produce sold by supermarkets in Nicaragua is still small, these findings indicate the
challenge to development programs to upgrade small farmers to have the capacity to
participate in restructuring market channels, and to find alternatives for the majority who
are unable to make the grade to participate in the new markets. Although this topic is
beyond the scope of this dissertation, a sustainable market oriented model for poor small

farmers is needed.

117

 

NOTES

' Note that in September 2005 Ahold sold its stake to Wal-mart. .

2 See Bardhan (1981) and Eswaran and Kotwal (1985) for theoretical treatment of these phenomena, and
Swinnen (2004) for comparative illustrations of similar inter-linkages in Central and Eastern Europe.

3 Such as in the case of Tate & Lyle global sugar company providing similar assistance to its farmers in
Slovakia to resolve “idiosyncratically” (for its specific group of suppliers) the market failures those farmers
faced in the technical assistance and credit and input markets; see Cow and Swinnen (2001)

4 There are three possible reasons why La Colonia wholesalers do not require the producers in their
preferred supplier lists to sort the tomatoes and only buy the top quality as Hortifruti does. The first is that
those producers are located in more distant regions and so their availability of alternative markets for
second quality tomatoes might generate important disincentives for producers, which indeed can generate
moral hazard problems (selling the whole production to other markets). Second, some of the wholesalers’
clients do not require only first quality tomatoes, which do not generate costly rejections downstream but
additional gains. Third, by buying all the production they increase their capacities to arrange collateral
payment for the potential risk generated of credit defaults of the producers in their preferred supplier lists.
In some way, they make arrangements to maintain long term relations with the inputs supplier companies
that provide credit to their farmers.

5 We acknowledge that the traditional wholesale market existed long before either the term supply chain or
the actions usually taken to be supply chain management existed, however we have used the name “supply
chain” to reflect that it represents the center of traditional flow of tomatoes from producers to consumers.

6 Note that the weights only take three values: wj=l= (ql/pl); wj=2=(q2/p2); wj=3=(q3/p3). The estimator
requires the true population proportions be known: (I) q.= population number of producers selling to CSU
supply chain over total population targeted (43/369); (2) q2= population of producers selling to La Colonia
supply chain over total population (26/369); and (3) q3= population of producers selling to traditional
markets over total population target (76/369). Sample proportions to be known (I) p. = sample of producers
selling to CSU supply chain over total sample (43/145); (2) p2 = sample of producers selling to La Colonia
supply chain over total sample (26/145); and (3) p3 = sample of producers selling to traditional markets
over total sample (76/145).

7 In order to test explanatory differences between production functional forms two F tests were performed
for the whole sample (as unfortunately our small sample does not.let us performing this test by market
channels). In the first test, the restricted form is represented by the Cobb Douglas production function and
the unrestricted form is the quadratic production function. The F statistic gave a value of 2.06. In the
second test, the restricted form is represented by the Cobb Douglas production function and the unrestricted
form is the translog production function. The F statistic equals 1.67. The F value of 99% significance level
is equal to 2.0938, hence we fail to reject in both cases the null hypothesis of the join equality of all
parameters between the production functional forms, which means that there are no explanatory differences
between the Cobb Douglas and quadratic and translog production functions.

3 The non-labor variable inputs were calculated by multiplying the quantities (standardized to Kilograms)
of each specific type of inputs times its price (standardized to Kilograms) and then aggregated by variable
category.

9 The three production functions are estimated using an OLS procedure. We found no evidence of
heteroskedasticity after performing the Breusch-Pagan test.

'0 The F statistic for the Chow Test between Producers selling to CSU and La Colonia is 14.65. It is equal
to 6.48 for the Chow Test between the producers that sell to the CSU and Traditional channels and equal to
3.92 between producers selling to La Colonia and traditional market channels. The F value of 99%

118

 

significance level is equal to 3.67, hence we reject in all cases the null hypothesis of the join equality of all
parameters between the production functions.

” As economic efficiency is the result of both technical efficiency and allocative efficiency we look at
both. Note that marginal physical productivity (MPP) is calculated by multiplying the input's market
channel production function parameter times the market channel sample average output (tomato
production) divided by the market channel input value sample aVerage.

Subsequently, we calculated the input's marginal product value (MPV) by multiplying the MPP times the
sample average tomato price.

'2 Land is the value in USS of acquiring one hectare calculated from the full sample and presented as
average for each supply chain in Table 7.

'3 Variable non-labor inputs include: fertilizers, chemicals foliars, insecticides, fungicides, herbicides
aggregated by costs of all types.

'4 We are assuming that spot markets are unregulated, which is perhaps not always the case.
'5 Note that the local bank LAFISE-Bancentro is co-owner of the large-scale processor “MACESA”.

‘6 A farmer selling to [CI is part of the supermarket supply chain, and those selling to the auction market
are part of the traditional markets. Farmers in the large-scale processor supply chain are those that don’t
sell to ICI or to auction markets.

'7 Compensatory growth is defined as a phase of rapid grth following food restriction
'8 See above how the supply chains are defined.

'9 In the first stage, because we collected the data in a two stage procedure (first purposive, second stage
random) using an endogenous stratification approach we estimate the multinomial logit model using the
Weighted Estimation Sampling Maximum Likelihood (WESML), developed by Manski & Lerman (1977).

20 To estimate our model we use a two stage Heckman procedure approach, where in the first stage we
estimate the market channel participation using multinomial logit model and in the second stage we
estimate cattle supplies by using a seemingly unrelated regression (SUR) system controlling for selection
bias of unobservable characteristics in the first stage (producers skills, risk) and conditional probability of
being in a market channel.

2' The price includes all cattle sales by category (female and male) over the last 2004 production season.
Producers were asked to recall the average price at which they sold their cattle. Variation in price might be
due to time variations within 2004.

22 We calculate from the first stage multinomial model the Inverse Mills Ratios (IMR) to correct for
selection bias'as suggested by following the standard procedures such as in Dubin and McFadden (1984)
and Bourguignon et al. (2004) in order to estimate in the second stage the unbiased marketing supply
equations using the Zellner’s seemingly unrelated regression (SURE).

119

 

REFERENCES

Aadland, D. & Bailey, D. (2001). Short-Run Supply Responses in the US. Beef-Cattle
Industry. American Journal of Agricultural Economics, 83 (4), 826-839.

Alvarado, I. and Charmel, K. (2002). The Rapid Rise of Supermarkets in Costa Rica:
Impact on Horticultural Markets. Development Policy Review, 20 (4), 473-485.

Bardhan, PK. (1991). A Note on Interlinked Rural Economic Arrangements. In Bardhan,
P. (ed.) The Economic Theory of Agrarian Institutions, Oxford: Oxford University
Press.

Berdegué, J ., Balsevich F., Flores L., and T. Reardon. (2005). The Rise of Supermarkets
in Central America: Implications for Private Standards for Quality and Safety of
Fruits and Vegetables. Food Policy, 30(3): 254-269.

Berdegué, J. A., Ramirez, E., and Faiguenbaum, S. (2005). Fomento de la pequefia
agricultura en Chile: Ideas para el debate pziblico. Serie Debates y Temas Agrarios 5.
Rimisp: Santiago, Chile.

Bellemare M. F. and C. B. Barrett, (2004). An Ordered Tobit Model of Market
Participation: Evidence from Kenya and Ethiopia. American Journal of Agricultural
Economics.

Bourguignon, F., F ournier, M. & Gurgand, M. (2004). Selection Bias Corrections Based
on the Multinomial Logit Model: Monte-Carlo Comparisons. Département Et
Laboratoire D ’e'conomie Théorique Et Applique’e, DELTA. Working paper No 2004-
20. Paris, France.

Campbell, R. (2000). Analysis of the Nicaraguan Meat Processing Industry. Chemonics
International Inc. Report for USAID, Managua, Nicaragua.

Carter, MR. and K.D. Wiebe. (1990). Access to Capital and Its Impact on the Agrarian
Structure and Productivity in Kenya. American Journal of A gricultural Economics,
72(5):]146-1150.

Codron, J .M., E. Giraud-Héraud and LG. Soler, (2005). Minimum quality standards,
premium private labels, and European meat and fresh produce retailing. Food Policy,
30(3), 270-283.

120

 

CORFOGA, 2002. Estudio de Mercado de Habitos de Consumo de la Carne.
Corporacion Ganadera (CORFOGA). Costa Rica. http://www.corfoga.org, accessed
March 14, 2005.

Cosslett, S. R. (1981). Efficient Estimation of Discrete-Choice Models. Structural
Analysis of Discrete Data. Cambridge, Massachussetts: MIT Press.

Cosslett, S. R. (1981). Maximum Likelihood Estimator for Choice Based Samples.
Econometrica, 49(5): 1289-1316.

Dubin, J .A. & D.L. McFadden. (1984). An Econometric Analysis of Residential
Appliance Holdings and Consumption. Econometrica, 52 (2), 345-362.

Eswaran, M. and A. Kotwal (1985). A Theory of Contractual Structure in
Agriculture. American Economic Review 75: 353-367.

Greene, W.H. (2000). Econometric Analysis. New Jersey: Prentice Hall.

Goetz, SJ. (1992). A Selectivity Model of Household Food Marketing behavior in Sub-
Saharan Africa. American Journal of Agricultural Economics, Vol. 74 (2), 444-452.

Heckman, J. (1979). Sample Selection Bias as a Specification Error. Econometrica,
47(1):]53-162.

Henson, S. and T. Reardon (2005). Private agri-food standards: Implications for food
policy and the agri-food system. Food Policy Review, 30(3), 241-253.

Hobbs, J. (1997). Measuring the Importance of Transaction Costs in Cattle marketing.
American Journal of Agricultural Economics, 79 (4), ’1083-1095.

Holloway, G., C. B. Barrett and S. Ehui, (2004). Bayesian estimation of the double hurdle
model In The Presence Of Fixed Costs. Journal of International Agricultural Trade
and Development..

Hueth, B.M., Ligon, E., Wolf. S. and W. Wu. (1999). “Incentive Instruments in
Agricultural Contracts: Input Control, Monitoring, Quality Measurement, and Price
Risk,” Review of Agricultural Economics, 21 (2): 374-389.

IICA-PROVIA, (2002). Estudio de la cadena de comercializacion de la carne bovina.
Instituto Interamericano de C ooperacio'n para la A gricultura (IICA). Mimeo.
Managua, Nicaragua.

' 121

 

Jarvis, S.L. (1974). Cattle as Capital Goods and Ranchers as Portfolio Managers: An
Application to the Argentine Cattle Sector. The Journal of Political Economy, 82 (3),
489-520.

Key, N. and D. Runsten. 1999. “Contract Farming, Smallholders, and Rural Development
in Latin America: The Organization of Agroprocessin’g Firms and the Scale of
Outgrower Production.” World Development 27 (February): 381—401.

Key, N., Sadoulet, E. and de Janvry, A. (2000). Transactions Costs and Agricultural
Household Supply Response. American Journal of Agricultural Economics, Vol. 82
(2), 245-259.

McFadden, D. (1976). “Quantal Choice Analysis: A Survey”. Annals of Economic and
Social Measurement, 5 (1976), 363-390.

Manski, F. C. & S.R. Lerman (1977). The Estimation of Choice Probabilities from
Choice Based Samples. Econometrica, 45 (8), 1977-1988.

Lau, L]. and P. Yotopoulos. (1971). A Test for Relative Efficiency and Application to
Indian Agriculture,” American Economic Review, 61(1), March: 94-109.

Leiva, E. F. (2004). Nicaragua Retail Food Sector Nicaraguan Retail Food Sector 2004.
Global Agricultural Information Network Report # NU4001. USDA Foreign
Agricultural Service. Washington DC: USDA.

Manski, F. C. and SR. Lerman. (1977). The Estimation of Choice Probabilities from
Choice Based Samples. Econometrica, 45 (8), 1977-1988.

Ospina, E. & Shumway, CR. (1979); Disaggregated Analysis of Short-Run Beef Supply
Response. Western Journal of Agricultural Economics 4 (2), 43-59

Paarsch, H.J. (1985). Micro-Economic Models of Beef Supply. The Canadian Journal of
Economics, 18 (3), 636-651.

Pe'rez, E. (2004). El Acceso de pequefios y medianos Ganaderos de Carne a Mercados
Dindmicos: El caso de Costa Rica. Mimeo, International Livestock Research Institute.
Managua, Nicaragua.

Peterson, H.C., A. Wysocki, and SB. Harsh. (2001). Strategic choice along the vertical
coordination continuum. International Food and Agribusiness Management Review, 4
(2001), 149-166.

122

 

Pingali, P.L. & Rosegrant, M.W. (1995). Agricultural commercialization and
diversification: processes and policies. Food Policy, 20 (3), 171-185.

Pomareda, C., Brenes, E. and L. Figueroa (1997). La' Industria de La Ganaderia de Carne
Bovina en Nicaragua: Condiciones de Competitividad. CEN 541. Centro
Latinoamericano de Competitividad y Desarrollo Sostenible (CLA CDS) del INCAE.
San José, Costa Rica.

Pomareda, C. (2004). Adjustments in the livestock sector to Supermarkets’ Strategies:
The case of Costa Rica. Paper presented at FAO International Seminar on Structural
Change in the Livestock Sector: Social and Environmental Implications for Policy
Making. Bangkok, Thailand.

Precios Mensuales de la Carne por Minorista y Mayorista (2003, 2004). Sistema de
Informacion de Precios y Mercados (SIPMA). Ministerio Agropecuario y F orestal de
la Republica de Nicaragua (MA GFOR). http://www.magfor.gob.ni; accessed March
14, 2005.

Reardon, T. & Berdegue, J .A. (2002). The Rapid Rise of Supermarkets in Latin America:
Challenges and Opportunities for Development. Development Policy Review, 20(4),
317-34.

Rosen, S., Murphy, K.M. & Scheinkman, J .A. (1994). Cattle Cycles. The Journal of
Political Economy, 102 (3), 468-492.

Sadoulet, E. and A. De Janvry (1995), Quantitative Development Policy Analysis, The
John Hopkins University Press, Baltimore and London.

Schejtman, A. (1996) “Agroindustry and small-scale agriculture: conceptual guidelines
for a policy to encourage linkage between them.” Report LC/R.1660, Economic
Commission for Latin America and the Caribbean, Santiago, Chile.

SIDE. (2004). La Experiencia de las Subastas Ganaderas en Costa Rica: Beneficios para
los Pequefios Ganaderos. Servicios Internacionales para el Desarrollo Empresarial
S.A. (SIDE). San José, Costa Rica.

Spreen, T.H., Bruce, A.M. & White, T.K. (1980). Modeling the Cattle Subsector: A
Demonstration in Guyana. Canadian Journal of Agricultural Economics, 28 (1), 17-
32.

123

 

Swinnen, J. (2004). Vertical Integration, Interlinking Markets, and Growth in Transition
Agriculture. Paper presented at the European Association of Agricultural Economics
Congress in Ghent on “Role of Institutions in Rural Policies and Agricultural
Markets.” September. -

Thrupp, LA. (1995). Bittersweet Harvest for Global Supermarkets: Challenges in Latin
America ’s Agricultural Export Boom, Washington: World Resources Institute.

Vella, F. (1998). Estimating Models with Sample Selection Bias: A Survey. Journal of
Human Resources 33 (1), 127-169.

Williamson, O.E. (1973). Markets and hierarchies: some elementary considerations. American
Economic Review 63, 316-325.

124

   

llllllllllllllllfllljlllllllllllll