theoretical and empirical analysis of the economics of traceability adoption in food supply chains

THEORETICAL AND EMPIRICAL ANALYSIS OF THE ECONOMICS OF TRACEABILITY ADOPTION IN FOOD SUPPLY CHAINS A Dissertation Presented by DIOGO M.. In the second model, along with generating a pre

THEORETICAL AND EMPIRICAL ANALYSIS OF THE ECONOMICS OF

TRACEABILITY ADOPTION IN FOOD SUPPLY CHAINS

A Dissertation Presented

by DIOGO M SOUZA MONTEIRO

Submitted to the Graduate School of the University of Massachusetts Amherst in partial fulfillment

of the requirements for the degree of DOCTOR OF PHILOSOPHY

February 2007 Resource Economics

UMI Number: 3254897

3254897 2007

Copyright 2007 by Souza Monteiro, Diogo M.

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346 All rights reserved.

by ProQuest Information and Learning Company

© Copyright by Diogo M Souza Monteiro 2007

All Rights Reserved

THEORETICAL AND EMPIRICAL ANALYSIS OF THE ECONOMICS OF

TRACEABILITY ADOPTION IN FOOD SUPPLY CHAINS

A Dissertation Presented

by DIOGO M SOUZA MONTEIRO

Approved as to style and content by:

Department of Resource Economics

To Marta for her love, patience, and courage throughout these years

To my parents, José and Madalena for all the sacrifices they made for me, and to my

siblings, Madalena and Francisco for their unconditional support

To my uncle Rafael Monjardino for his friendship and encouragement

To the memory of my grandparents

To my great-grandfather Prof Jorge Monjardino and to my uncle Prof Fernando de

Oliveira Pinto whose academic careers inspire and challenge mine

To all my friends in Portugal, especially to Ezequiel, José Maria, Macário, and Pedro

“For this I was born, and for this came I into the world; that I should give testimony to the truth Every one that is of the truth,

hears my voice Pilate said to him: What is the truth?”

John 18, 37-38

“Deus quer, o homem sonha, a obra nasce.”

Fernando Pessoa, Mensagem

ACKNOWLEDGMENTS

First and foremost I would like to thank my advisor, Professor Julie A Caswell, for encouraging me to apply to the Resource Economics doctoral program at the

University of Massachusetts Amherst, and for patiently guiding, mentoring and

supporting me throughout these years Julie has not only been an excellent advisor and a role model, but also a true friend My wife Marta and I have an enormous debt of

gratitude to Julie and to Richard for helping us to adjust to a new environment and for always being available for any professional or personal issues we faced during this time

I am also very grateful and obliged to the members of my committee: Professors Daniel A Lass, Anna Nagurney, and John K Stranlund for their excellent comments and very insightful suggestions, and for taking time to be part of this research It was a great privilege and honor for me to collaborate with such a fine and renowned group of University of Massachusetts faculty members I learned a lot from each of you and hope

to have opportunities for future collaboration

I want to thank the Fundação para a Ciência e a Tecnologia (Portuguese

Foundation for Science and Technology) for funding this research and providing for travel expenses to seminars and meetings where preliminary versions of this work were presented I also have to thank Henrique Curto, António Fragata, Nelson Isidoro,

Armando Torres Paulo, Sergio Pereira, Catarina Ribeiro, Domingos Santos, and João Miguel Silva for their generosity in granting me access to data on traceability adoption

in the Portuguese pear industry, without which the empirical chapter of this thesis could not be done

My research was influenced by several people each indirectly contributing to part of the final product I am particularly in debt to Dr Neal Hooker who encouraged

me to think about multi-ingredient foods, to Professor Barry Field for his support and help in thinking more in economic terms, and especially to Professor Joe Moffitt Even though Professor Moffitt was not part of my committee he nevertheless contributed very kindly and generously to my research, helping me sharpen my initial ideas and always showing great confidence in my ability to carry on my work

I had the enormous fortune of having a great group of fellow graduate students

in my years at the University of Massachusetts Each of them enriched and broadened

my personality I would especially like to thank José Alves, Sven Anders, Katherine Boschert, Seda Erdem, Dmytro Matsypura, David McEvoy, Philip Mellizo, Sirisha

Naidu, Linus Nyiwul, Michael Schuppli, Richard Volpe, Lava Yadav, and Yi (Alan)

Zhong for their friendship and encouragement, and for walking along with me through the hurdles of graduate school

Margaret A Cialek, Wendy J Curtice, and Barbara A Talenda also deserve recognition for all their help, kindness and patience in dealing with the administration and bureaucracies of the University and for attending to special requirements Elisa Campbell from OIT and Carol Foster from the Graduate School generously helped me with the final manuscript styles and formatting

Last, but certainly not least, a very special thank you must go to Father Richard Cleary S.J., to the University of Massachusetts Amherst Newman Center Community and to the Portuguese graduate students Catholic group in Boston who contributed to

my sanity, sense of reality, and growth as a human being

ABSTRACT

THEORETICAL AND EMPIRICAL ANALYSIS OF THE ECONOMICS OF

TRACEABILITY ADOPTION IN FOOD SUPPLY CHAINS

FEBRUARY 2007 DIOGO M SOUZA MONTEIRO, B.S., UNIVERSITY OF EVORA

M.S., MEDITERRANEAN AGRONOMIC INSTITUTE OF ZARAGOZA

M.S., UNIVERSITY OF MASSACHUSETTS AMHERST Ph.D., UNIVERSITY OF MASSACHUSETTS AMHERST

Directed by: Professor Julie A Caswell

Traceability systems are increasingly implemented in food supply chains to mitigate food safety hazards and to improve information management and logistics While traceability systems are largely voluntary and driven by consumer demand in the United States (US), in the European Union (EU) they are enforced by regulations and were implemented to restore consumer trust in the food supply Traceability systems establish the path of information on food origins, attributes, and production and

processing technologies from farm to fork, thus increasing transparency in the food chain

As the consumption of pre-prepared foods increases around the world, new types of food safety hazards may arise Previously independent food supply chains may now come together in facilities producing multi-ingredient, pre-prepared food products Implementing traceability in multi-ingredient food chains may mitigate food safety risks

This research presents three essays, two theoretical and one empirical, analyzing the economics of traceability adoption in food supply chains and at the farm The first essay investigates under what conditions voluntary or mandatory traceability systems are preferred in single ingredient supply chains The second addresses the conditions for full, partial or no traceability in multi-ingredient food chains Finally, the third essay analyzes what has influenced traceability adoption at the farm level in the Portuguese pear industry An overall aim of this research is to examine how network effects impact the levels of traceability flowing along single- and multi-ingredient food supply chains

This dissertation contributes to the economics of traceability in food supply chains showing that mandatory traceability may be inevitable if there are large public benefits from traceability, liability rules are not perfectly imposed, and monitoring and enforcement is effective Full traceability is feasible in multi-ingredient supply chains However partial traceability is perhaps a more realistic scenario as it may result in considerable savings for firms along the food chain Producer organization, retailer, and farm characteristics clearly influence the decision to adopt traceability Finally, network effects have a positive impact on the levels of traceability but a negative impact on the value of premiums

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS vi

ABSTRACT viii

LIST OF TABLES xii

LIST OF FIGURES xiii

CHAPTER 1 EXECUTIVE SUMMARY 1

2 MANDATORY AND VOLUNTARY TRACEABILITY ADOPTION IN SINGLE-INGREDIENT FOOD CHAINS 9

2.1 Introduction 9

2.2 Motivations for voluntary and mandatory traceability 12

2.3 Modeling traceability with network, agency, and vertical control theories 17

2.4 Models 21

2.4.1 Strictly private voluntary provision of traceability 25

2.4.2 Liability loss mitigation provision of traceability 31

2.4.3 Mandatory provision of traceability 34

2.4.4 Contract structure 39

2.5 Results 40

2.5.1 Comparing strictly private voluntary, liability and mandatory levels of traceability 40

2.5.2 Network effects on level of traceability and premium for the first tier firm 42

2.6 Conclusions and future research 46

3 TRACEABILITY IN MULTI-INGREDIENT FOOD SUPPLY CHAINS 54

3.1 Introduction 54

3.2 Operations research and economic approaches to traceability analysis 56

3.3 A Model of voluntary traceability in a multi-ingredient supply chain 61

3.3.1 Assumptions 62

3.3.2 Model 66

3.4 Results and discussion 74

3.5 Conclusions and future research 82

4 TRACEABILITY ADOPTION AT THE FARM LEVEL: AN EMPIRICAL ANALYSIS OF THE PORTUGUESE PEAR INDUSTRY 87

4.1 Introduction 87

4.2 Motivations for traceability adoption 89

4.3 Modeling traceability adoption 93

4.3.1 Theoretical and discrete choice models of traceability adoption 94

4.3.2 Approach to empirical estimation 98

4.4 The Portuguese pear industry 101

4.5 Survey methodology and data 103

4.6 Factors affecting traceability adoption in the Portuguese pear industry 106

4.7 Conclusions and future research 110

APPENDICES A SURVEY FOR EMPIRICAL STUDY – PORTUGUESE VERSION 119

B SURVEY FOR EMPIRICAL STUDY – ENGLISH VERSION 123

BIBLIOGRAPHY 127

LIST OF TABLES

2.1 Levels of traceability and premium for each firm 50

2.2 Implicit levels of private, liability, and voluntary traceability 51

2.3 Network effects on levels of and premiums for traceability 52

3.1 Traceability levels and premiums along the supply chain 86

4.1 Effects of premium, farmer’s profile, and size of farm on optimal traceability 114

4.2 Descriptive statistics for survey variables (N=138 valid cases) 115

4.3 Maximum and penalized likelihood estimates for traceability adoption 116

4.4 Cross tabulation of the variables traceability adoption and sales to the UK 117

4.5 Penalized likelihood estimates for sales to the UK (model 3) and traceability adoption (model 4) 118

LIST OF FIGURES

2.1 Network structure of a single-ingredient food traceability system 49

2.2 Efficient scale traceability level 53

3.1 Network structure of a multi-ingredient food traceability system 85

4.1 Profile likelihood for sales to UK by β estimates 113

CHAPTER 1 EXECUTIVE SUMMARY

Since 1998, in the aftermath of a sequence of food safety scares in the United Kingdom and other European countries, traceability began to be discussed as a policy choice to recover the trust of European Union (EU) consumers in food safety and quality Traceability systems establish the path of information on food origins,

attributes, and production and processing technologies from farm to fork, thus

increasing transparency in the food chain

Beef identification and traceability from farm to fork became mandatory in the

EU through European Commission and Parliament regulations 1760/2000 and

1825/2000 Then regulation 178/2002 of the European Parliament and of the Council on general food law, imposed mandatory traceability for all food products marketed in the

EU It entered into force in January 2005 Thus while in 2000 mandatory traceability was only imposed in one single-ingredient supply chain, namely the beef industry, the new EU food laws of 2002 extended its application to all foods

An increased interest in food safety information and the corresponding

enhancement of traceability systems raises at least two types of issues at the global level One concerns policy making and another industry development The first arises from different perceptions and orientations toward food safety policy in different parts

of the world, for example in the United States (US) and EU The second derives from the globalization of business and the increasing sophistication of food manufacturing and supply chains, particularly those involving pre-prepared foods

In contrast to the EU, in the US mandatory traceability is not generally

considered an acceptable policy choice to mitigate food safety hazards, perhaps because the reaction of consumers to food scares has not been as dramatic Golan et al (2003) argue that mandatory traceability systems will be too costly and unnecessary for product differentiation, especially if specific attributes of value to consumers are not targeted They also assert that traceability systems may be inefficient for the purpose of

increasing food system safety Analyzing the demand for traceability, Dickinson and Bailey (2002), and Hobbs et al (2005) found that North American consumers are

willing to pay a premium for traceability, especially if it is supported by a quality

assurance system

Clearly the EU and the US have different perspectives regarding the usefulness

of traceability and the way it should be implemented in the food industry An

implication of this is a likely increase in disputes at the World Trade Organization as the EU effectively imposes traceability on all food This justifies closer scrutiny of the implications and economic value of different traceability systems

As the consumption of pre-prepared foods, which often involve the combination

of different ingredients following a recipe, increases around the world new types of food safety hazards may arise This is because previously independent food supply chains now come together in the facility producing the multi-ingredient food This may increase risks of food safety hazards due to cross contamination Implementing

traceability in multi-ingredient food chains may mitigate food safety risks Trinekens and Beulens (2001), Hobbs (2002), Hofstede (2002), Meuwissen, et al (2003), and Van der Vorst (2004) made significant contributions to the understanding of traceability

adoption in single ingredient food chains but research on the implementation of

traceability in multi-ingredient chains is still scarce

Golan et al (2004) surveyed several different systems of traceability in US food industries concluding that in the US traceability tends to be solely motivated by economic incentives A study of voluntary traceability adoption in the European meat and poultry sectors found that its main drivers were higher production uncertainty, higher chances of moral hazard and opportunistic behavior, increasing monitoring costs, and inability to identify food traits (Buhr 2003) Recently, empirical studies by Mora and Menozzi (2005), focusing on the cost components and drivers of traceability, and

agro-by Banterle, Stranieri, and Baldi (2006) using a transactions costs perspective on

traceability, found that its adoption is less costly when firms are larger and have

previously implemented quality assurance systems These studies also indicated that traceability contributed to the reduction of internal and external product failures, and is

a strategic asset facilitating contractual relations, reinforcing trust, and improving the ability to assign liability There is no formal empirical analysis looking at what

motivates traceability adoption at the farm level However, this is a critical issue

because if farmers do not implement traceability it becomes impossible to have it from farm to fork

From an economic perspective, the key question is what are the main drivers and dynamics of traceability systems? To address this question, this research presents two theoretical essays analyzing the economics of traceability and an empirical study of traceability adoption at the farm level The first theoretical essay investigates the

conditions in which voluntary or mandatory traceability systems are preferred in single

ingredient supply chains The second theoretical essay discusses the conditions for full, partial or no traceability in multi-ingredient food chains The empirical essay analyzes what has influenced traceability adoption at the farm level in the Portuguese pear industry An overall aim of this research is to examine how network effects impact the levels of traceability flowing along single and multi-ingredient food supply chains

The second chapter of this dissertation presents the first theoretical essay It investigates the conditions for preferring voluntary versus mandatory traceability using three alternative models These combine network, agency, and vertical control theories

to study and compare implementation of a traceability system along a supply chain with three tiers In the models, traceability is a vertical and downstream flow of information governed by a sequence of contracts Downstream levels of traceability depend on upstream levels Thus, to have traceability downstream firms must set an appropriate premium to assure a firm immediately upstream accepts a contract for the provision of traceability In the first model the motivation for traceability is strictly voluntary and is

in response to consumer demand In the second model, along with generating a

premium paid by consumers, traceability adoption mitigates a stochastic liability loss due to food safety hazards Finally, the last model is of mandatory traceability, where in addition to being motivated by consumer willingness to pay and liability loss

mitigation, firms choose the socially optimal level of traceability in order to avoid a penalty set by a regulator In each of these models, the third tier firm downstream is the claimant of benefits from consumer willingness to pay, bears the stochastic liability losses of food safety hazards, and faces the regulatory penalty for not providing the socially optimal level of traceability Chapter two compares voluntary and mandatory

systems of traceability, and analyzes how network effects or externalities impact the optimal choice of traceability and their potential impacts in shaping policies

The main finding in chapter two is that leaving the decision to adopt traceability entirely up to the private sector may lead to an under provision of traceability, which may impede effective prevention and mitigation of food safety risks Mandatory

traceability, defined as the requirement to provide the socially optimal level of

traceability, should only be considered if food safety hazards have a high probability of occurrence, involve considerable private and social losses, and can be easily monitored and enforced The network effects have opposite influences depending on whether they impact levels of traceability or their corresponding premiums Effects on traceability levels due to changes in the marginal premium paid by consumers, in the probability of food safety hazards, and in the external benefits of traceability to society are all

positive The impacts on premiums set in the contract with the first tier firm will be zero

or negative if average costs are represented by U-shaped curves and the level of

traceability required is equal to or below the efficient scale level

The second theoretical essay, presented in chapter three, investigates the

voluntary implementation of full, partial, or no multi-ingredient traceability in a tiered supply chain with firm asymmetries The modeling strategy uses network,

three-agency, and vertical control theories In this model, there are several firms in both the first and second tiers, each linked with only one firm upstream As in chapter two, traceability is a vertical and downstream flow of information on product origins,

attributes, and processing technologies However, to implement traceability the third tier firm, producing a multi-ingredient food, must propose a contract to at least one of

its ingredient suppliers in the second tier Following the liability model of the second chapter, the third tier firm in the multi-ingredient supply chain adopts traceability to receive a premium from consumers and to mitigate a stochastic liability loss in the event

of a food safety hazard

The main conclusion on the multi-ingredient supply chain traceability model is that full traceability is feasible However partial traceability is perhaps a more realistic scenario Partial traceability may result in considerable savings and be more rational, as some of the ingredients composing the multi-ingredient food may have very limited probabilities of food safety hazards or enter in small proportions in the final product, which may result in lower benefits than costs from traceability

Another conclusion is that both horizontal and vertical network effects have to

be considered when analyzing the implementation of traceability in multi-ingredient supply chains The results show that horizontal network effects are always positive implying that traceability levels are complements across the ingredients composing the output of the third tier firm Consequently the response of the third tier firm to demand for more traceability may be very inelastic and only occur if it is for a reasonable

number of ingredients Vertical network effects result from changes in traceability premiums paid by consumers to the third tier firm and in probabilities of food safety hazards The results show that vertical effects are positive; as consumers raise their willingness to pay for traceability or the probabilities of food safety hazards increase, more traceability is demanded upstream What is not clear is if this change makes firms along the chain better or worse, which will depend on the structure of traceability costs and how benefits are distributed

Finally, chapter four presents an empirical analysis of traceability adoption at the farm level in the Portuguese pear industry This industry is suited for analysis of traceability adoption because farmers appear to adopt different levels of traceability, depending on their characteristics, location, affiliation with farmer organizations, and marketing strategy Farmer associations play a very important role in this region,

organizing most marketing activities, streamlining pear production, and helping farmers

to adopt quality assurance systems such as EurepGAP (Euro-Retailer Produce Working Group Good Agricultural Practices), for which traceability is a major requirement The dataset contains 138 observations on pear producers affiliated with six farmer

organizations in the “Oeste” (western) region of Portugal This region produces about

80 percent of the total annual Portuguese pear crop The decision to adopt the

EurepGAP traceability level is analyzed through a discrete choice model and estimated with the penalized likelihood estimator This estimation technique was recently

developed by Heinze and Schemper (2002) to deal with the monotone likelihood or quasi-separation problem It is particularly suited to correct the small sample bias frequently encountered in the estimation of discrete choice models

The results show that traceability adoption at the farm level is best described by

a two-step decision process Farmers first select the market they wish to sell to and then the level of traceability, according to the market requirements Larger farms, those complying with quality assurance systems such as a Protected Designation of Origin (PDO), and full time farmers affiliated with larger producer associations have higher odds of adopting more stringent traceability standards such as the EurepGAP scheme

On the other hand, higher productivity levels, lower levels of education, and location in regional centers of production reduce the odds of EurepGAP traceability adoption

These findings have important implications for public policy and firm strategic decision making In the region analyzed, traceability was adopted in response to a mix

of private and public incentives Industry leaders indicated that although traceability adoption involved considerable investments, it is starting to pay off because it can be used to improve the way information is managed The results show that public

authorities may wish to cooperate with producer organizations in deciding how to promote traceability adoption at the farm level For example, economies of scale in information collection and management as well as avoidance of network externalities may occur if farmer associations decide to invest in a traceability system that holds the registries for all their affiliates

This dissertation contributes to understanding of the economics of traceability

by showing that mandatory traceability may be inevitable if there are large public benefits from traceability, liability rules are not perfectly imposed, and monitoring and enforcement is effective Further, full traceability is feasible in multi-ingredient supply chains, but partial traceability saves costs along the supply chain Membership in producer organizations, retailer demands, and farm characteristics influence the

decision to adopt traceability in the Portuguese pear industry Finally, network effects have a positive impact of the levels of traceability However, network effects on

premiums paid to firms upstream in the supply chain depend on the structure of costs and on how benefits are distributed along the chain

CHAPTER 2 MANDATORY AND VOLUNTARY TRACEABILITY ADOPTION IN

SINGLE-INGREDIENT FOOD CHAINS

2.1 Introduction

Food supply chains are complex networks of interdependent firms, involved in the production, processing, transportation, marketing and sales of foods from farms to consumers In the past decades the agri-food system became increasingly knitted

together and, with globalization of food markets, distances between agricultural

production and consumption became larger This occurred not only in terms of physical distances but also in the number of tiers in the supply chain These fast paced changes

in food supply are unprecedented in human history and their consequences to society are yet to be fully understood One of the most striking elements of the new

environment is the increasing need for coordination, demonstrated by the growing number of contract agreements in food chains (MacDonald et al 2004; James, Klein, and Sykuta 2005) Another is the concern with the level of information and

transparency on food quality and safety, origin, and production and processing

technologies (Caswell 2006; European Parliament and Council 2002) Amongst other motivations and advantages, contracts and more information may reduce uncertainty and improve risk management, or shift it more efficiently along food chains, thus

reducing transaction and food safety costs

There has been notable progress in agricultural and food technology, production, and processing procedures Both private and public sectors have invested heavily to improve food safety standards through new technologies, management practices and regulations A prime example of this is the adoption of mandatory hazard analysis and

critical control points (HACCP) However, regardless of all these efforts old safety threats still prevail and new hazards are appearing every year, justifying further

investigation of responses to existing interventions and consideration of alternative or complementary instruments

Voluntary and mandatory food traceability systems are increasingly used to improve information sharing across food supply chains There is still considerable debate at the domestic and international level on whether traceability is an appropriate tool to improve food safety and, moreover, if mandatory traceability is justified Simply put traceability is an information system through which firms in different tiers of a supply chain share details on product origins, attributes, and production and processing technologies Traceability should not be seen as a panacea for food safety problems, rather it is a tool to improve information management along food supply chains It may

be seen more as a complement than as a substitute to existing public and private

interventions to improve food quality and safety

The objective of this chapter is to contribute to the debate on the economics of traceability It investigates under what conditions voluntary or mandatory traceability systems are preferred when there is a need for more information in a single ingredient supply chain Another goal is to investigate how and in which direction network

externalities affect optimal levels of traceability and respective premiums

The European Union (EU) was particularly affected by a series of food scares in the early 1990's leading to consumer distrust in farmers, agri-business firms, retailers, and public authorities Similarly the US, Japan, and Canada had their share of food safety incidents but, with the exception of Japan, it seems that their impact at the

consumer level was not as widespread as in the EU This succession of events led private and public authorities to propose further interventions aimed at restoring

confidence in the food supply

Information asymmetries and imperfections are important sources of food safety and quality distrust throughout the supply chain and at the consumer level Henson and Trail (1993) suggest that information remedies, along with process standards, product performance standards, and pecuniary measures could correct or mitigate food safety market failures However, they point out that provision of information may itself be imperfect, due to the nature of foodborne risks, the public nature of information, and asymmetries in the supply of food safety or quality information

Starting in 2000, through European Commission and Parliament regulations 1760/2000 and 1825/2000, beef identification and traceability from farm to fork became mandatory in the EU Since January 2005, through regulation 178/2002 on general food law, traceability became mandatory for all food products marketed in the EU More recently, the European Parliament and the Council regulation 1830/2003 imposes traceability for products containing at least 0.9 percent of genetically modified

materials Traceability is thus a pivotal element in new EU food legislation and a tool to mitigate food safety hazards Contrasting with this view and policy choice, Golan et al (2003) argue that mandatory traceability systems will be too costly and unnecessary for product differentiation, especially if specific attributes of value to consumers are not targeted They also argue that such systems may be inefficient for the purpose of

increasing food system safety, as they could reduce the incentives for firms to innovate

in their processes in order to improve safety levels

Traceability implies a flow of information between firms located in different tiers of a supply chain; each of these obtains information upstream but also creates new pieces of information There may be both dependencies and complementarities between pieces of information introduced in a traceability system covering the entire supply chain Complementarities may lead to network effects and thus impact the level of traceability adoption This research draws on vertical coordination, network, and agency theories to develop an integrated framework for analyzing the adoption of traceability as

a coordination problem between firms along food supply chains

2.2 Motivations for voluntary and mandatory traceability

Traceability is becoming an important instrument supporting existing food quality and safety assurance systems in agri-food supply chains worldwide The idea of gathering and sharing information from producer to consumer is not new or exclusive to the food industry For example, traceability is used in the mailing and package

expediting industries, allowing customers to track their packages from the point of origin to its destination In the airlines industry, tracking systems are used to make sure luggage and passengers are on board the same plane and go to the same destination Traceability is required by the Federal Aviation Administration to ensure that only approved and certified parts are used on aircraft production and maintenance (US Department of Transportation 1998 and 1999)

While traceability in the mail and packaging industries is voluntary and

motivated by consumer demand, in aviation it is mandatory and motivated by the need

to assure flight safety A relevant question is what level of circumstances motivates

the nature and consequences of risk and failures are much larger when building and repairing airplanes Also, the structures of these industries are different Finally,

building and repairing planes is much more complex than shipping mail or packages In the case of the food industry, there are no clear cut answers for this question of

voluntary versus mandatory traceability, as within this industry there is a wide variety

of cases, some resembling the mailing industry, while others are similar to aircraft manufacturing and maintenance Nevertheless, it appears that the larger the risks and their consequences, as well as the complexity of food supply chains, the more

appropriate would be mandatory traceability

Golan et al (2004) studied United States traceability systems implemented in the grain and oilseed, fresh produce, and cattle and beef industries They found that these systems are voluntary and efficient levels of traceability are determined weighing private costs and benefits Dickinson and Bailey (2002) and Hobbs et al (2005) found that North American consumers are willing to pay a premium for traceability, especially

if it is supported by a quality assurance system The private sector developed several types of procedures and instruments to avoid market failures arising when private levels

of traceability do not correspond to the social optimum (Golan et al 2004) Tort liability regarding food safety hazards may be another important driver of traceability In the US and Europe, courts may take into account the fact that an offending firm has traceability and reduce the amount of the penalty Thus there are at least two motivations to

voluntarily adopt traceability: rent seeking due to consumer demand and mitigation of exposure to court ruled liability for food safety hazards Golan et al (2004) argue that

unless there is clear evidence of market failures, voluntary traceability should be

preferred

Antle (1995) argues that as with other forms of market intervention, mandatory traceability is only justified if markets fail to provide efficient and effective levels of information to prevent or minimize food safety hazards and if the net benefits of

interventions are positive Mandatory traceability systems have been in place for beef since 2000 in the European Union and Japan (Souza Monteiro and Caswell 2004) However, imposing traceability is costly; the USDA estimates a total cost of $500 million in the US over a period of six years for implementing a traceability system from cattle farms to slaughterhouses (Bailey, Robb, and Checketts 2005) This cost should be compared to possible benefits of having mandatory traceability, among which could be regaining and maintaining full access to international markets such as Japan, to which

US exports totaled $1.64 billion in 2000 (US Department of Agriculture 2003)

In voluntary traceability systems, firms have flexibility regarding whether to adopt traceability and at what level The above discussion suggests that traceability should only be adopted if there is a clear private benefit to firms Firms may voluntarily adopt traceability because consumers are willing to pay a premium for it, it may

improve logistic efficiencies, it may facilitate product differentiation, and it may

mitigate liability losses A disadvantage of voluntary systems is that since not all firms adopt traceability, part of the market remains anonymous Another drawback is that firms may adopt a lower level of traceability than what is needed to facilitate the

prevention of food hazards

Information introduced in traceability systems has public and private

components and there may be asymmetric information along the supply chain (Henson and Trail 1993) Traceability systems may be based on credence information related to food quality and safety (Crespi and Marette 2001) Verbeke (2005) suggests there may

be inconsistencies between consumer demand for information and its provision by producers, processors, or retailers He also observes that traceability systems record and transmit information on a wide variety of attributes and not all consumers are able to understand the cues provided, which may reduce their willingness to pay for

traceability Hence the private provision of traceability may not fully account for its social benefits, which may lead to a socially inefficient provision of traceability In a recent paper, Pouliot and Sumner (2006) study traceability in the context of liability for food safety hazards They show how its adoption increases incentives for private firms

to provide safer food They also demonstrate that as traceability levels increase in an industry, the incentives to free-ride decrease and, even with a large number of firms, the probability that food is safe will rise Cho and Hooker (2006) compare performance and process standards imposed by a regulator with a limited budget who aims to reduce risk effectively Contrary to the general wisdom, the authors show that process are preferred

to performance standards if input use is highly variable, the industry has a majority of inefficient firms and the regulator pays less attention to variability of industry-level compliance and more to the standard level

Mandatory traceability systems may compel all agents involved in the market to share information with their partners They may also force the provision of the socially optimal level of information that takes into account the public value The voluntary

provision of traceability may be too low because firms fail to internalize the public benefits of traceability to a society, which may justify public intervention imposing mandatory traceability A benevolent regulator should not only analyze different

options of public intervention but also consider whether for each policy choice social benefits are larger than social costs

The literature supports arguments that either voluntary or mandatory traceability may be appropriate in different circumstances Incentives to adopt traceability may also

be influenced by the liability rule applied to the food system In the US strict liability is used, meaning that a seller will be held responsible even if it is not aware of the hazards

of its product (Cooter 1991) In the EU, the new general food law (Regulation (EC) N°178/2002 of the European parliament and of the Council, January 2002) prescribes the use of classic liability, through which operators are only liable for actions and occurrences under their control (Standing Committee on the Food Chain and Animal Health 2004) The US liability rule is more stringent than that of the EU, which may explain the lower support for traceability in the US food industries

In economics, liability rules are often seen as ex-post policies, whereas

regulations are ex-ante policy options to correct externalities Ex-post interventions are seen as a substitute for ex-ante policies and economists tend to favor the least costly

option (Kolstad, Ulen, and Johnson 1990) However, there are many instances in which

both ex-ante and ex-post policy options are used; a recent example is the above cited

new EU general food law Furthermore, Kolstad, Ulen, and Johnson (1990) argue that both these policy options may be inefficient when regulators have imperfect

information, when law suits are not issued, when injurers can easily apply for

bankruptcy, or when there is uncertainty over the legal standard defined by judges

Under these circumstances there may be complementarities in the use of both ex-ante and ex-post policies

A closer analysis of the type of traceability adopted and its impact throughout the supply chain can shed light on the circumstances that support adoption of voluntary versus mandatory traceability

2.3 Modeling traceability with network, agency, and vertical control theories

Traceability systems can be envisioned as networks, where firms in different tiers are nodes linked by phone, fax, or digital connections through which information flows on product origin, attributes, and production technologies However, a traceability system does not emerge spontaneously; rather it involves a coordinated effort and typically requires leadership from a firm in the chain or from an external party, for example a regulator This coordination can be governed by contractual relations, as in agency theory where a principal needs an agent to take an action on its behalf and for its benefit The definition of traceability proposed by the EU implies a flow of information between firms along a supply chain Information must be gathered or created at each tier and information downstream depends on information acquired upstream Thus the level of traceability of a firm downstream includes traceability from upstream This is similar to the case analyzed with vertical control models, as in Royer (1998) where firms downstream employ output from firms upstream to produce a final product

The literature on supply chains and networks is very relevant as a framework on which to build an integrated perspective of information imperfections There are many different streams in the network theory literature In the context of traceability, the

industrial organization and institutional economics perspectives, along with those operations research and management science, are particularly relevant In the first perspective the term network relates more to a good or a commodity than to firms in supply chains In their seminal paper, Katz and Shapiro (1985) mention products for which the utility of consumption rises with the number of consumers using them In this literature a network externality is defined as a variation in the value of a good

associated with changes in the number of its users Economides (1996) suggests that network externalities require coordination or compatibility between firms producing network goods Traceability may be subject to network externalities because the more firms in a supply chain are willing to adopt it the more complete might be the level of information and the value of traceability in the chain

When modeling supply chains using an operations research perspective, the objective is to define an optimal flow for each link between the nodes representing locations of firms in the supply chain In management science methodologies such as variational inequalities, described by Nagurney (1999), have been applied to solve highly complex problems involving several different agents linked by different paths in multiple-tiered supply chains

The relevance of the industrial organization approach to network economics for the present research is that it emphasizes the need to consider coordination between firms in the food chain and how network effects impact levels of traceability in the chain The operations research and management science approaches provide a powerful mathematical framework through which the supply chain can be graphically represented and analyzed as an integrated system of independent agents

Menard (2004) offers an alternative view of a network, which he defines as

“(…) all arrangements defining the set of recurrent contractual ties among autonomous entities” (p 348) Clearly this puts network in the context of institutional economics and agency theory Coase (1937) was one of the first economists to point out the differences between market transactions and those occurring within organizations such as firms He observed that transaction costs should be considered when a firm or a manager therein

is deciding whether it is more economical to use an organization, where resources are allocated by the authority of a person, or a market, which uses the price mechanism As

he later explained, it is the comparison between the costs incurred to operate a firm and those of using the market that ultimately influence the decision to start or expand a firm,

as these directly impact profitability (Coase 1988) The insights of Williamson (1989) are also very relevant to the analysis of traceability adoption in food chains, as

transaction costs and asset specific investments must be taken into account when

developing such a system Williamson places transaction costs at the core of the

analysis of organizational economics and proposes the use of a contractual approach Vertical integration is an application of transaction cost theory, for it involves the choice of using the market, keeping firms more independent in the supply chain, or merging down- or upstream

Agency theory, as presented by Coase and Williamson, is relevant for

traceability modeling as it provides alternative settings in which to base models and provide guidance on how to think about the tradeoffs between traceability systems based on markets or within organizations

Royer (1998) proposes different models to investigate vertical coordination in the food industry In his base model a processor B purchases inputs from producer A, and produces output using a fixed proportions technology with constant returns to scale Although Royer aims to explain the effects of vertical coordination on output, prices, and the welfare of various market participants, the models proposed are suited to study traceability along supply chains Here Royer’s model of vertical integration is used as a basis to explain how information flows between partners along the supply chain

Hornibrook and Fearne (2001) propose an application of organizational

economics theory to model multi-tiered supply chains They analyze perceived risk in the UK beef supply chain The beef supply chain is represented by a pair wise series of principal-agent relationships Apart from the agents in the initial and terminal node of the chain, all other elements play both principal and agent roles An important feature of this model is that agents have to consider actions taken at previous dyads and perceived risk in their contracting decisions In the models proposed in this chapter, the second tier firm also has a double role as a principal and an agent; and firms downstream must take into account the levels of traceability provided from upstream firms

Meuwissen et al (2003) consider three possible traceability systems in the European beef industry which are relevant to the structure of the supply chain used in the models proposed in this paper The first system, type A, is one where each firm in the food chain only retains its own information In type B traceability design,

downstream firms store and transfer their own information along with that gathered upstream Finally, in the third system, type C, each firm in the supply chain sends

information to an external party that stores information and assures information is traced along the supply chain

To model traceability adoption in single ingredient supply chains, network theory is useful to describe the direction of information flows between firms, to offer a graphical representation of the supply chain, and to discuss network externalities or effects Then agency theory can be used to analyze how a set of contracts are written between firms located in different tiers to govern the provision of traceability along the chain Finally, since a traceability system implies that downstream levels of traceability depend on the existence of traceability from upstream, the flow of information along the chain can be suitably approached with a vertical control model

2.4 Models

The analysis of voluntary and mandatory traceability adoption in single

ingredient supply chains is based on three models In the first model, seen as the

baseline, the motivation for traceability adoption is consumer’s willingness to pay The second model explores the case where the motivation for traceability adoption is the mitigation of liability losses due to stochastic food safety hazards Finally, the third model analyzes a mandatory regulation to enforce the socially optimal level of

traceability, in which the public good dimension of information is considered The following key assumptions are common to all models in order to enable comparisons between them

Assumption 1: The supply chain has three tiers or stages as in the model of vertical control proposed by Royer and Bhuyan (1995) In the first tier (upstream) a firm produces a commodity, in the second tier a processor purchases output from the

first tier firm and processes or assembles it into an intermediate good Finally, in the third tier (downstream) a final output processor or a retailer uses the output from the second tier firm, transforms it, and sells it to consumers Associated with this product flow there is a downstream flow of information on origin, characteristics, and

technologies used at each stage of the transformation process occurring along a supply chain This is what Pape at al (2005) define as a food processing traceability scheme,

on which the attributes and transformation of inputs into a final product can be

completely listed at each tier of production

Information flows are governed by contracts proposed sequentially by the third tier firm to the assembler in the second tier, and by this assembler to the commodity producer in the first tier Each contract defines levels of traceability, denoted by γ1and a

monetary compensation (p i , i = 1,2,3 ), where the subscript indicates the tier in the supply chain where each firm is located

The third tier firms can be seen as the leader or principal of this supply chain This is consistent with Hennessy, Roosen, and Miranowski (2001) who assume that

processors in the US food industry and large retailers chains in the EU are de facto

leaders of the supply chain and have a “pilot role in coordinating identity preservation and in the provision of information to customers” (p 870) The second tier firm plays a double role, as it is an agent with respect to the third tier firm but a principal regarding the first tier

Figure 2.1 shows a graphical representation of this supply chain Starting at the second tier, each firm records and passes down its own information along with that

obtained upstream, which is consistent with the “type B” traceability system described

by Meuwissen et al (2003)

Assumption 2: All firms in the supply chain maximize profits Along with fixed net benefits from selling output, which is independent of traceability levels, each firm may get additional revenue from taking a contract to provide traceability downstream

or, in the case of the third tier firm, accept the premium to offer traceability to

consumers

Assumption 3: Traceability is assumed to be a continuous variable, as it

represents the amount of information stored and transmitted between firms along the supply chain, regardless of its type or format Furthermore, there is a chain relation between levels of traceability downstream and upstream summarized in the following functional relations: γ2 =γ γ2( )1 and γ3 =γ γ3( )2 Thus the level of traceability

downstream depends on that gathered upstream These functions are assumed to be linearly increasing in all arguments (i.e., γi′ > and0 γi′′ = , i = 2, 3) Since, by 0

definition, traceability requires a flow of information across each tier of the supply chain, this is not possible unless second and third tier firms gather information

upstream, thus (0) 0γi = , i=2, 3

Assumption 4: Adopting traceability is costly Information must be gathered,

perhaps involving expensive laboratory procedures; registered in paper or digital formats; and exchanged through email, websites, fax, phone or other forms Case studies by Giacomini, Mancini, and Mora (2002); Van der Vorst (2004), and Gellynck

et al (2005) show that traceability implementation requires large initial investments, mostly for information technologies Operating the system also involves fixed and

variable costs Thus, assume costs are given by ( )c i γi , which are increasing and strictly convex As Sodano and Verneau (2003) argue, traceability can be thought of as an information production technology and “usually processes of information production exhibit high fixed costs and a typical U-shaped average variable costs curve” (p 4) Assuming U-shaped average costs will facilitate the interpretation of comparative statics analysis in the results section

Assumption 5: It is important to distinguish between levels of information

exchanged between firms in the supply chain and the information needed to design contracts Information flowing amongst firms in a traceability system relates to food safety and quality attributes, origin, and/or processing technologies Examples include the farm where cattle are born, the type of pesticide used to control apple diseases, or whether cheese is produced with pasteurized milk These types of information are designated as traceability In agency theory, assumptions on information are crucial for contract design and this is the other sense in which the word information is used here

To avoid misunderstanding, information on transaction costs is called contract

information The models below assume that traceable information is always true in order to abstract from the problem of adverse selection Contract information is perfect,

in the sense that costs and benefits of information to each firm in the supply chain are common knowledge

In the following sections these assumptions are used to formulate three models

of traceability adoption in food chains Additional assumptions, specific to each model, will be introduced as the models are presented below

2.4.1 Strictly private voluntary provision of traceability

This section presents the first model of voluntary traceability Following

Dickinson and Bailey (2002) and Hobbs et al (2005), the implementation of traceability

is motivated by consumer willingness to pay a linearly increasing premium (p3 3γ ) to the third tier firm This firm has to offer information on product origin, attributes, and production and processing technologies Then, given the structure of the supply chain described in assumption 1, and the functional relation between traceability levels in assumption 3, the third tier sets a price (p2) on the contract with the second tier firm Likewise, the second tier firm sets a premium (p1) for the level of traceability

contracted with the first tier firm

The first tier firm makes a fixed profit (π1) from selling a commodity This is the same profit the firm gets if it decides to sell in a market that does not require

traceability To this output profit it may add a marginal premium (p1) per unit of

information (γ1) if it accepts a contract from the second tier firm to adopt traceability This will only occur if the firm receives at least its reservation profit Traceability is costly These costs are denoted as c1( )γ1 , which by assumption 4 is an increasing and convex function of traceability The problem of first tier firms is:

The second tier plays a critical role in the traceability system If it fails to hire the participation of a first tier firm, by assumption 3 above, it cannot produce its own information and traceability will not be possible (γ2 =γ γ2( )1 and γ2(0) 0= ) The third tier offers a contract to the second tier; if this firm accepts it then it must propose a contract to the first tier firm

As with the first tier firm, the second tier firm can sell in a spot market in which traceability is not required with a profit π2 When the second tier firm takes a contract from the third tier firm, a premium (p2) is added per unit of traceability information (γ2) To assure participation of the first tier firm in a contract for traceability, the

second tier firm must offer the reservation profit (π1) which is the profit from selling output in the spot market i.e.:

1 1

1 1 1

γ

γ γγ

constraint, the second is the incentive compatibility constraint and can be seen as a penalty for failing to deliver the correct level of traceability