"t‘ k. n.

111

«‘In...)

 

 

am.“
‘1:
\ ,
V ‘ .1
.1-

‘ .93"?! a
Rang“;
“Ulc-

. I .
. 1.34 ”(N a,
. [imfiiunnu‘knz

1
.4
.ti »
nun r.

.5 v2 J41
.innfififih. ax
‘ rmul': it.”

{at , . .
2&4). 6.8% ‘ , A .uma‘
6:4 that-.d;|.t:% . , t! R
. 4.» f. . {m‘

at. .fiqhw...

fawn ...L.mm.$

tr... 43:35.1.

.V. IL $51-34. ‘5.“
£1. 5....

15.x. .A 212.51%.
Vfit‘rv OI. “bah-I‘u' ”it ‘
l .. 2|. “flux: Hum.“ afa-

.. .: L\
40: “Hurltl:
5!. 1 304389.13...)
1’31»: 2? an}. w:
nlnnrmm.
Li!” firm ,
321...; u: “wmhflfinj .2.
at...» E... .i .4. W4...“ hail...
a. st!

K tau... .0”

is .L v: 3.! ts
: It‘s.
. 1., {thin . no
.JllniuAfi «nu-.4... . I. “Ink
:1 El V . 4
.1..U,.&lla i1.n.§.l..lt.~)..
.. allzoctf. 15.02.}: .
(‘9! d) 1117.“!
zit-:95 505533.. at
.(iu‘l‘fit‘k)
. . 3.!!! .
xi .

p.23"

.7?- “ .

.L‘fi.-I\.r l\ ,1-
...k .1 l\
. n . u v 1'

an? .. . 7 . .

 

THESIS

g? LIBRARY
NM Michigan State
University

This is to certify that the
thesis entitled

COW-CALF PRODUCER PREFERENCES FOR
VOLUNTARY TRACEABILITY SYSTEMS AND SYSTEM
ATTRIBUTES

presented by

Lee L. Schulz

has been accepted towards fulfillment
of the requirements for the

degree in Agricultural Economics

 

 

m MMflL

' Méjor Professor’ 5 Signature

July 25, 2008

 

Date

MSU is an affinnative-action, equal-opportunity employer

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:/Prolecc&Pres/ClRC/DateDue‘indd

 

COW-CALF PRODUCER PREFERENCES FOR VOLUNTARY TRACEABILITY
SYSTEMS AND SYSTEM ATTRIBUTES

By

Lee L. Schulz

A THESIS

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

MASTER OF SCIENCE
Department of Agricultural Economics

2008

ABSTRACT

COW-CALF PRODUCER PREFERENCES FOR VOLUNTARY TRACEABILITY
SYSTEMS AND SYSTEM ATTRIBUTES

By

Lee L. Schulz

Substantial losses can occur if animal identification systems cannot quickly and
adequately identify individual animals, the premises where they were located, and their
movements throughout production and processing. This creates a need met by this study
in determining how traceability systems should be designed and promoted in order to
improve voluntary participation rate. This research utilized a survey of US. cow-calf
producers to identify cow-calf producer preferences and perceptions regarding voluntary
traceability systems and system attributes and in turn determined what type of voluntary
traceability systems would receive the greatestsupport. Meeting this core objective
allowed for better identification of the potential success of alternative voluntary
traceability systems that could exist in the beef industry. A second tier of research
questions included examinations of mandatory vs. voluntary NAIS preferences, self-
revelation of current NAIS participation, and the most current concerns and important
issues to cow-calf producers regarding traceability. Results have policy implications as
the optimal voluntary traceability system hinges critically upon cow-calf producer
perceptions of traceability systems and system attributes. Results indicate the importance
of considering producers’ perceptions and preferences regarding traceability when

designing traceability systems.

Copyright by
LEE L. SCHULZ
2008

Dedicated to my Mom and Dad

iv

ACKNOWLEDGEMENTS

I would like to thank the Department of Agriculture, Food and Resource
Economics at Michigan Stateifor providing me with many opportunities. Furthermore, I
want to give a general thanks to all the faculty and students who I have interacted with
who have been so helpful and supportive throughout my time at MSU.

Special thanks go to my major professor, Dr. Glynn Tonsor, for his friendship and
guidance. Glynn has played a crucial role in my development as a graduate student and
researcher. This work would not have been possible without his contributions. 1 am truly
grateful for his help and am appreciative that he has gone above and beyond to help me
succeed. I would also like to thank the members of my thesis committee, Dr. Roy Black,
Dr. Chris Wolf, and Dr. Dan Buskirk. They have always provided information and
insights that have led to higher quality output.

Furthermore, I would like to thank Joleen Hadrich and Nicole Olynk for
continually answering my questions and for their help throughout my studies and
research. Their experience and expertise has proven to be indispensable assets
throughout my time at MSU. I cannot thank Nicole enough for her assistance in the
distribution of the survey for this thesis. Joleen and Nicole have proven to be great
friends and have been very influential in my success.

Finally, I would like to thank my parents, Cliff and Carol, sister, Megan, and

girlfriend, Kelly, for all their support and encouragement.

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... VIII
LIST OF FIGURES ........................................................................................................ XIII
CHAPTER 1 : INTRODUCTION ....................................................................................... 1
1.1 PROBLEM STATEMENT .......................................................................................... 1
1.2 SCOPE ................................................................................................................... 3
1.3 ORGANIZATION OF THESIS .................................................................................... 5
CHAPTER 2: LITERATURE REVIEW ............................................................................ 6
2.1 TRACEABILITY SYSTEM DEFINITION ..................................................................... 6
2.1.1 Meat Traceability ............................................................................................ 8
2.1.2 Live Animal Traceability .............................................................................. 10

2.2 ECONOMICS AND CURRENT STATUS OF TRACEABILITY SYSTEMS ....................... 13
2.2.1 Economics of Traceability Systems .............................................................. 13
2.2.2 Current Status of Traceability Systems ......................................................... 16

2.3 CONSUMER IMPACTS/PREFERENCES ................................................................... 20
2.4 WORK NEEDED ................................................................................................... 23
CHAPTER 3: OBJECTIVES ............................................................................................ 25
CHAPTER 4: SURVEY AND DATA DESCRIPTION .................................................. 32
4.1 SURVEY DESIGN ................................................................................................. 32
4.2 SURVEY DATA .................................................................................................... 38
CHAPTER 5: CONCEPTUAL FRAMEWORK AND MODELS ................................... 42
5.1 FORECASTING FRAMEWORK ............................................................................... 43
5.1.1 Tobit Model .................................................................................................. 43

5.2 PROBIT MODEL SPECIFICATIONS ......................................................................... 45
5.2.1 Multinomial Probit Model ............................................................................ 45
5.2.2 Binary Probit ................................................................................................. 46
5.2.3 Endogeneity Bias .......................................................................................... 48
5.2.4 Trinomial Probit ............................................................................................ 50
5.2.5 Order Probit .................................................................................................. 50

5.3 CHOICE MODEL SPECIFICATIONS ........................................................................ 51
5.3.1 Multinomial Logit ......................................................................................... 52
5.3.2 Random Parameter Logit .............................................................................. 54
5.3.3 Latent Classification Model .......................................................................... 56
CHAPTER 6: EMPIRICAL MODELING ....................................................................... 58
6.1 TOBIT MODELS ................................................................................................... 59
6.2 PROBIT MODELS ................................................................................................. 61
6.2.1 Binary Probit ................................................................................................. 62

vi

6.2.2 Endogeneity Evaluation ................................................................................ 62

6.2.3 Trinomial Probit ............................................................................................ 62
6.2.4 Ordered Probit ............................................................................................... 64
6.3 MULTINOMIAL LOGIT MODELS ........................................................................... 65
6.4 RANDOM PARAMETER MODELS .......................................................................... 68
6.5 LATENT CLASSIFICATION MODELS ..................................................................... 69
6.6 WILLINGNESS-TO-ACCEPT ESTIMATES ............................................................... 70
6.7 WELFARE MEASURE ........................................................................................... 72
CHAPER 7: RESULTS .................................................................................................... 74
7.1 PRODUCERS’ PRACTICES AND PERCEPTIONS REGARDING TRACEABILITY ........... 74
7.1.1 Producers’ Forecast Analysis ........................................................................ 75
7.1.2 Registration in NAIS ..................................................................................... 78
7.1.3 Mandatory NAIS Beliefs .............................................................................. 80
7.1.4 Important Issues when Implementing Traceability Systems ........................ 83
7.1.5 Producer Concerns When Implementing Traceability Systems ................... 88
7.1.6 Issues in the Beef Industry when Implementing Traceability Systems ........ 95
7.2 CHOICE EXPERIMENT ANALYSIS ......................................................................... 96
7.2.1 Multinomial Logit Model ............................................................................. 96
7.2.2 Random Parameter Logit Model ................................................................. 101
7.2.3 Latent Classification Model ........................................................................ 106
7.3 PRODUCER WELFARE EFFECTS ......................................................................... 1 13
7.4 IMPLICATIONS ................................................................................................... 116
CHAPTER 8: CONCLUSIONS ..................................................................................... 136
APPENDICES ................................................................................................................ 139
APPENDIX 1: COW-CALF PRODUCER SURVEY ................................................ 140
APPENDIX 2: VARIABLES USED IN REGRESIONS AND DIAGNOSTICS ....... 146
APPENDIX 3: COW-CALF PRODUCER SURVEY DEMOGRAPHIC,
PERCEPTION, AND CHOICE EXPERIMENT SUMMARY TABLES ........ 148
3.1 Demographics ............................................................................................. 148
3 .2 Perceptions .................................................................................................. l 56
3.3 Choice Experiment ...................................................................................... 162
APPENDIX 4: SUPPLEMENTAL REGRESSIONS AND DIAGNOSTICS ............ 168
APPENDIX 5: SAS, STATA, AND LIMDEP CODE ................................................ 175
5.1 SAS ............................................................................................................. 175
5.2 STATA ........................................................................................................ 194
5.3 LIMDEP ...................................................................................................... 195
REFERENCES ............................................................................................................... 198

vii

LIST OF TABLES

Table 7.1. Producers’ Forecasts ........................................................................................ 77
Table 7.2. NAIS Premise Registration .............................................................................. 79
Table 7.3. Producers’ Beliefs Concerning Mandatory NAIS ........................................... 81
Table 7.4. MB to Monitoring/Managing Disease ............................................................. 85
Table 7.5. MB to Maintaining Current Foreign Markets .................................................. 86
Table 7.6. ME to Accessing Foreign Markets .................................................................. 86
Table 7.7. MB to Increasing Consumer Confidence ......................................................... 87
Table 7.8. MB to Concerns Regarding System Cost ........................................................ 89
Table 7.9. MB to Concerns Regarding System Liability .................................................. 90
Table 7.10. ME to Concerns Regarding Confidentiality .................................................. 91
Table 7.11. ME to Concerns Regarding Reliability .......................................................... 92
Table 7.12. ME to Concerns Regarding Non-Participating Firms Benefiting .................. 93
Table 7.13. ME to Concerns Regarding System Failures ................................................. 94
Table 7.14. ME to Implementation of COOL ................................................................... 95
Table 7.15. Multinomial Logit Estimates ......................................................................... 97
Table 7.16. Multinomial Logit Willingness-to-Accept ..................................................... 99
Table 7.17. Random Parameter Logit Estimates ............................................................ 101
Table 7.18. Random Parameter Logit Willingness-to-Accept ........................................ 105
Table 7.19. Latent Class Model Estimates ...................................................................... 108
Table 7.20. Latent Class Model Willingness-to-Accept ................................................. 111
Table 7.21. Welfare Effects ............................................................................................ 114
Table 7.1a. Producers’ Forecasts .................................................................................... 122

viii

Table 7.2a. NAIS Premise Registration Estimates ......................................................... 123

Table 7.3a. Producers’s Beliefs Concerning Mandatory NAIS Estimates ..................... 124
Table 7.4a. Monitoring/Managing Disease Estimates .................................................... 125
Table 7.5a. Maintaining Current Foreign Markets Estimates ......................................... 126
Table 7.6a. Accessing Foreign Markets Estimates ......................................................... 127
Table 7.7a. Increasing Consumer Confidence Estimates ................................................ 128
Table 7.8a. Concerns Regarding System Cost Estimates ............................................... 129
Table 7.9a. Concerns Regarding System Liability Estimates ......................................... 130
Table 7.10a. Concerns Regarding Confidentiality Estimates ......................................... 131
Table 7.11a. Concerns Regarding System Reliability Estimates .................................... 132
Table 7.12a. Concerns Regarding Free-Riding Estimates .............................................. 133
Table 7.13a. Concerns Regarding System Failures Estimates ........................................ 134
Table 7.14a. Implementation of COOL Estimates .......................................................... 135
Table A.3.1. Producers’ Gender (Survey Question 1) .................................................... 148
Table A.3.2. Producers’ Age (Survey Question 2) ......................................................... 148
Table A.3.3. Producers’ State of Residence (Survey Question 3) .................................. 148
Table A.3.4. States within the ERS Farm Production Regions ...................................... 149
Table A.3.5. ERS Farm Production Regions .................................................................. 149
Table A.3.6. States within the US. Regions ................................................................... 150
Table A.3.7. U.S. Regions .............................................................................................. 150
Table A.3.8. Farm Organizations of Membership (Survey Question 4) ......................... 150
Table A.3.9. Educational Description (Survey Question 5) ........................................... 151
Table A.3.10. Years Raising Cattle (Survey Question 6) ............................................... 152

ix

Table A.3.11. Expected Future Years Raising Cattle (Survey Question 7) .................... 152
Table A.3. 12. Estimated Annual Pre-tax Household Income (Survey Question 8) ....... 152
Table A.3.13. Off-Farm Household Income (Survey Question 9) ................................. 152
Table A.3.14. Labor Supplied by Paid Employees (Survey Question 10) ..................... 153
Table A.3.15. Feed Needs Produced by Own Farm (Survey Question 11) .................... 153
Table A.3.16. Production Claims (Survey Question 12) ................................................ 153
Table A.3.17. Marketing Methods (Survey Question 13) .............................................. 154
Table A.3.18. Beef Cows that Calved (Survey Question 14) ......................................... 154
Table A.3.19. Operation’s Sales at Production Stages (Survey Question 15) ................ 154
Table A.3.20. Market Finished Cattle and Own Feedlot (Survey Question 15) ............. 154
Table A.3.21. Current Animal Identification Methods (Survey Question 16) ............... 155
Table A.3.22. Current Premise Registration in NAIS (Survey Question 22) ................. 155
Table A.3.23. Issues in Designing a Traceability System (Survey Question 17) ........... 156
Table A.3.23a. Monitoring/Managing Disease (Survey Question 17) ........................... 156
Table A.3.23b. Increasing Consumer Confidence (Survey Question 17) ....................... 156
Table A.3.23e. Enhancing Marketability (Survey Question 17) .................................... 156
Table A.3.23d. Maintaining Current Foreign Markets (Survey Question 17) ................ 157
Table A.3.23e. Accessing Foreign Markets (Survey Question 17) ................................ 157
Table A.3.23f. Improving On-F arm Management (Survey Question 17) ...................... 157
Table A.3.23 g. Managing the Supply Chain (Survey Question 17) ............................... 157
Table A.3.23b. Enhancing Food Safety (Survey Question 17) ....................................... 158
Table A.3.24. Concerns in Designing a Traceability System (Survey Question 18) ...... 158
Table A.3.24a. Cost to Participating Producer ............................................................... 158

Table A.3.24b. Confidentiality of Information ............................................................... 158

Table A.3.24c. Reliability of Technology ...................................................................... 159
Table A.3.24d. Liability to Participating Producer ......................................................... 159
Table A.3.24c. Non-Participating Firms Benefiting ....................................................... 159
Table A.3.24f. Failure of System to Meet Stated Goals ................................................. 159
Table A.3.25. Common Traceability Statements (Survey Question 19) ........................ 160
Table A.3.25a. “Is More Cost Effective for Larger Cow-Calf Operations” ................... 160
Table A.3.25b. “Results in More Liability for Cow-Calf Producers” ............................ 160
Table A.3.25c. “IS Unnecessary if COOL was Implemented Nationally” ..................... 160
Table A.3.25d. “As a Mandated System is Exaggerated” .............................................. 161
Table A.3.26. Allocation of Benefits with Traceability (Survey Question 21) .............. 161
Table A.3.27. Allocation of Costs with Traceability (Survey Question 21) ................... 161
Table A.3.28. Should NAIS be Mandatory (Survey Question 23) ................................. 161
Table A.3.29. Forecast of Premises Registered (Survey Question 24) ........................... 161
Table A.3.30. Forecast of Premises Registered and RFID Use (Survey Question 24)... 161
Table A.3.31. General Choice Experiment Statistics ..................................................... 162
Table A.3.32. Complete Choice Experiments ................................................................ 162
Table A.3.33. Traceability Scenarios Block A ............................................................... 163
Table A.3.34. Traceability Scenarios Block B ............................................................... 164
Table A.3.35. Traceability Scenarios Block C ............................................................... 165
Table A.3.36. Traceability Scenarios Block D ............................................................... 166
Table A.3.37. Traceability Scenarios Block E ................................................................ 167
Table A4 1. Managing the Supply Chain Estimates ...................................................... 168

xi

Table A.4.2. ME to Managing the Supply Chain ........................................................... 168

Table A.4.3. Enhancing Food Safety Estimates ............................................................. 169
Table A.4.4. ME to Enhancing Food Safety ................................................................... 169
Table A.4.5. Improving Marketability Estimates ........................................................... 170
Table A.4.6. ME to Improving Marketability ................................................................. 170
Table A.4.7. Improving On-Farm Management Estimates ............................................. 171
Table A.4.8. ME to Improving On-Farm Management .................................................. 171
Table A.4.9. Economies of Scale Estimates ................................................................... 172
Table A.4.10. ME to Economies of Scale ....................................................................... 172
Table A.4.11. Shifts in Liability Estimates ..................................................................... 173
Table A.4.12. ME to Shifts in Liability .......................................................................... 173
Table A.4.13. Mandatory Traceability Exaggeration Estimates ..................................... 174
Table A.4.14. ME to Mandatory Traceability Exaggeration .......................................... 174

xii

LIST OF FIGURES

Figure 1.1. Organization of Thesis ..................................................................................... 5
Figure 4.1. US. Regions ................................................................................................... 39
Figure 5.1. Conceptual Model Progression Flow Chart ................................................... 42

xiii

KEY TO SYMBOLS AND ABBREVIATIONS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abbreviation Definition
AIN Individual Animal Identification Number
APHIS Animal, Plant Health Inspection Service
BSE Bovine Spongiforrn Encephalopathy
COOL Country-of-Origin Labeling
CDF Cumulative Distribution Function
ERS Economic Research Service
F MD Foot and Mouth Disease
GAO Government Accountability Office
HACCP Hazard Analysis Critical Control Points
ID Identification
IIA Independence of Irrelevent Alternatives
IRB Institutional Review Board
LCM Latent Classification Model
ME Mean Marginal Effects
MLE Maximum Likelihood Estimation
MNL Multinomial Logit
NAIS National Animal Identification System
NIWP National Identification Work Plan
PIN Premises Identification Number
PVP Process Verified Program
QSA Quality Assessment Program
RF ID Radio Frequency Identification
RPL Random Parameter Logit
TTA Traceability, Transparency, and Enhanced Quality Assurances
UCRIHS University Committee on Research Involving Human Subjects
US. United States
USAIP United States Animal Identification Plan
USDA United States Department of Agriculture
WTA Willingness-to-Accept
WTP Willingness-to-Pay

 

xiv

 

CHAPTER 1: INTRODUCTION

1.1 Problem Statement

Substantial losses can occur if animal identification systems cannot quickly and
adequately identify individual animals, the premises where they were located, and their
movements throughout production and processing. Recent events both domestically and
internationally have identified the need for animal identification systems. One of these
events was the announcement on December 23, 2003 that a cow in the US. was
diagnosed with Bovine Spongiform Encephalopathy (BSE or Mad-Cow Disease). Even
though consumer demand in the United States (US) for beef products remained strong in
the weeks following the announcement, the US. beef industry and US. government
recognized the need for a traceability system that went beyond the current US. system
which was not designed to routinely track individual or groups of animals once they leave
a premise.

The National Identification Work Plan (N IWP) was the first official public effort
in the US. to examine the possible implementation of a US. animal identification (ID)
system. The NIWP was developed by a task force formed in April 2002 consisting of
over thirty livestock organizations and was coordinated through the National Institute for
Animal Agriculture. The working plan for the implementation of the animal ID system
as suggested by the NIWP was later called the US. Animal Identification Plan (USAIP)
in 2003. The USAIP called for the establishment of individual premises ID by the
summer of 2004, individual animal ID by 2005, and full implementation and compliance
(all covered species and their movements - both interstate and intrastate) by July 2006

(Bailey and Slade, 2004).

Many of the efforts for these initial goals saw opposition and encountered many
obstacles. Through this initial groundwork, the United States Department of Agriculture
(USDA), Animal, Plant Health Inspection Service (APHIS) has attempted to implement a
nationwide beef traceability system to help producers and animal health officials respond
quickly and effectively to animal disease outbreaks (e. g., Foot and Mouth Disease-F MD,
Bovine Spongiform Encephalopathy-BSE, etc.) and food recalls (e.g., ground beef due to
E. coli 0157:H7). With traceability taking the forefront of supply chain issues, the
reality is that traceability systems are imminent, whether voluntary as they currently are
or mandatory, which have been discussed within the industry. Many of the efforts of the
USAIP have evolved into the National Animal Identification System (N AIS) with the
only Significant difference being that NAIS is listed as “tech neutral” in its policies
relating to animal identification, meaning that NAIS eliminated radio frequency
identification (RFID) as the stated standard for animal ID. The National Animal
Identification System is the broadest and most comprehensive effort ever launched in the
US. to enhance the ability to quickly identify and contact animal premises, promote
animal identification, and develop animal movement and tracing capabilities (Schroeder
et al., 2007). Initial deadlines for full implementation and compliance have elapsed with
participation rates for the establishment of individual premises ID and individual animal
ID below expectation. This creates a need met by this study in determining how
traceability systems should be designed and promoted in order to improve voluntary
participation rate. Furthermore, a national livestock ID system can be used to launch

more extensive quality assurance programs.

In addition to animal health management, US. and international consumers
demand meat product safety assurances and they have revealed a willingness-to-pay
(WTP) for meat traceability (Dickinson and Bailey, 2002; Hobbs, 2003) and for attributes
that could easily be verified by traceability systems (Loureiro and Umberger, 2003).
Traceability is being used as the basis of competitive product differentiation strategies by
food firms seeking to assure consumers of the presence of credence attributes related to
production or processing methods (Hobbs, 2006) as well as by producers who seek
proactive information and quality verification throughout production.

Because cow-calf producers are the first player in the beef supply chain and vary
widely in scale and production practices of their operations, it is crucial to consider the
demographics and perceptions of cow-calf producers when attempting to implement
industry-wide programs/systems. This is especially important when attempting to
implement individual animal traceability and maximize participation rates of these
systems as the views of these producers will most certainly impact the success or failure

of these efforts.

1.2 Scope

One of the core objectives of this research was to determine preferences cow-calf
producers have for alternative voluntary traceability systems and their attributes.
Questions addressed include, how sensitive are producer preferences to price adjustments
(premiums/discounts), what are producer preferences for the entity maintaining data, and
what composition of additional advanced traceability information (e. g., age verification

vs. production practice information) maximizes expected voluntary participation rates?

Furthermore, how do producer demographics, perceptions, and current
production/technology practices affect each of these questions and are these impacts
homogeneous or heterogeneous across producers? A second tier of research questions
included examinations of mandatory vs. voluntary NAIS preferences, self-revelation of
current NAIS participation, and the most current concerns and important issues to cow-
calf producers regarding traceability. In addition, our analysis obtained producer
forecasts of expected voluntary participation rates, self-revelation of current participation,
and examination of other issues to improve government and industry efforts to further
increase voluntary traceability participation in the US. cow-calf industry.

Various methods were employed in model specification including: multinomial
Logit (MNL), random parameter Logit (RPL), and latent classification models (LCM) for
examining producer preferences for traceability participation, Probit models for
examining producer perceptions and current practices regarding traceability systems, and
Tobit models to identify the most optimistic and pessimistic of producers regarding future
voluntary participation rates. For each research question addressed, alternative model
specifications were used to identify the impact of producer demographics and
perceptions, while also considering producer heterogeneity.

Results have policy implications as the optimal voluntary traceability system
hinges critically upon cow-calf producer perceptions of traceability system attributes.
Overall, this research was able to provide valuable information for future policy
deliberations and may assist organizations in charge of program administration to better

manage resources used in animal ID practices.

1.3 Organization of Thesis

The thesis proceeds in the following manner: Chapter 2 reviews the literature
drawing reference from existing traceability systems, shows how traceability has and will
continue to be vital to the beef industry, and overviews methods used to analyze
preferences of cow-calf producers; Chapter 3 identifies the objectives of this research;
Chapter 4 gives a description of the cow-calf producer survey including the design and
data obtained; Chapter 5 provides the conceptual framework and models; Chapter 6
describes the empirical modeling to be used in estimation; Chapter 7 applies the
modeling techniques to the data to give results and policy implications; and Chapter 8

summarizes and concludes. Figure 1.1 is provided as a visual flow chart of the thesis.

Figure 1.1. Organization of Thesis

 

 

 

 

CHAPTER 1: INTRODUCTION 1) CHAPTER 2: LITERATURE
REVIEW

 

 

 

 

 

 

 

 

CHAPTER 4: SURVEY AND (1:: CHAPTER 3: OBJECTIVES
DATA DESCRIPTION

 

 

 

 

 

 

 

 

CHAPTER 5: CONCEPTUAL ,:(> CHAPTER 6: EMPIRICAL
FRAMEWORK AND MODELS MODELING

 

 

 

 

 

 

 

 

CHAPTER 8: CONCLUSIONS (7; CHAPTER 7: RESULTS

 

 

 

 

 

 

 

 

 

CHAPTER 2: LITERATURE REVIEW

The literature review provides information regarding traceability systems and the
methods used for determining cow-calf producer preferences for voluntary traceability
systems and attributes. The literature review is divided into for main sections: (1)
Traceability System Definition, (2) Economics and Current Status of Traceability

Systems, (3) Consumer Impacts/Preferences for Traceability, and (4) Work Needed.

2.1 Traceability System Definition

There is no international agreement or a “one-type-fits-all” definition of
traceability; however, past literature has provided various definitions of traceability.
Bulut and Lawrence (2007) provide a collection of traceability definitions. The
International Organization for Standardization (ISO 9001, 2000) defines traceability as
“the ability to trace the history, application, or location of an entity by means of recorded
identifications” (The Food Business Forum, 2005). Golan et al. (2004) found this
definition quite broad and suggests: “record keeping systems that are designed to track
the flow of product or product attributes through the production or supply chain”.
According to Smyth and Phillips (2002), the supply chain literature sees traceability as
“(an) information system necessary to provide history of products and services from
origin to the point of sale.” Mennecke et al. (2006) define traceability as “the ability to
retrieve the history, treatment, and location of the animal that a cut of meat comes from,
through a recordkeeping and an audit system or registered identification program” (Bulut,

and Lawrence, 2007).

More specifically within the beef supply chain, full ‘gate to plate’ or ‘farm to
fork’ traceability is an extensive form of beef traceability which provides the ability to
follow products forward from their source animal (i.e., birth or ancestry), through growth
and feeding, slaughter, processing, and distribution, to the point of sale or consumption
(or backward from the consumer to the source animal) (Becker, 2007). Particularly, a
two-part-system has developed in the beef and cattle industry; meat traceability and live
animal traceability (Bulut and Lawrence, 2007). Linking these two systems at the stage
of slaughter and processing is an ongoing challenge for the industry (Golan et al., 2004).

For this two-part system it is important to define certain aspects of traceability.
Bulut and Lawrence (2007) provide definitions of external traceability, internal
traceability, chain traceability, forward traceability, and backward traceability. External
traceability refers to traceability of product or product attributes through the successive
stages of production (e.g. cow-calf producer, auction barn, feedlot, slaughter, and
processing). Where as, internal traceability refers to traceability within the plant or
production unit, which may be a part of Hazard Analysis Critical Control Points
(HACCP) plans. External traceability may require some degree of internal traceability
(Lupin, 2006). Chain traceability refers to traceability throughout the entire food chain.
Backward traceability, traceback, or tracing is defined as “the ability to identify the origin
of a particular unit and/or batch of product located within the supply chain by reference
to records held upstream” (New Zealand Trade and Enterprise, 2006). Forward
traceability, traceforward, traceup, or tracking is defined as “the ability to follow the path

of a Specified unit of a product and/or batch through the supply chain as it moves between

organizations toward the final point-of—sale or point-of-service” (New Zealand Trade and
Enterprise, 2006).

Liddell and Bailey (2001) further distinguish traceability from transparency and
quality assurance notions. Transparency refers to the public availability of production
information at each stage of production and quality assurances refer to practices to ensure
food quality and safety, which could be intrinsic such as back fat and curing or extrinsic

such as animal welfare and environmental preservation (Bulut and Lawrence 2007).

2.1.1 Meat Traceability

Smith et al. (2000) define meat traceability as the ability to identify the origin of
animals or meat as far back in the production sequence as necessary to ascertain
ownership, identify parentage, assure safety and determine compliance in branded or
source-verified beef programs. A method commonly referred to as farm-to-retail
traceability is a system that maintains the identity of all cuts from the farm through the
cutting and distribution system. Farm-to-retail tracability is very expensive and
essentially requires new construction and extensive capital investment and data
infrastructure. This type of system is very rare although many consumers think that this
is the system in place for beef (Jensen and Hayes, 2006). A more common type of
traceability involves traceability from the farm-to-carcass. This type of system is termed
batch traceability. The life history of the animal is tracked for each carcass or primal cut,
but the ability to trace the animal parts through the cutting floor is lost. Instead, the meat
is cut and processed in batches. The final retail product can be traced to a particular batch

in the processing plant. This type of system is relatively inexpensive, especially if the

batches are large (Jensen and Hayes, 2006). For US. meat slaughter and processing
plants, traceability is currently voluntary beyond the record keeping required by Federal
Meat Inspection Act, Wholesome Meat Act, HACCP plans of 1996, and BSE regulations
of 2004 (Bulut and Lawrence, 2007).

Past literature provides definitions and descriptions of the roles and functions of
meat traceability. Sanderson and Hobbs (2006) have broadly defined, five roles of
traceability systems that include: (i) improve inventory and logistic management; (ii)
improve management of food recalls in the event of a food safety problem; (iii) limit the
broader (public) impacts of food safety and/or herd health problems; (iv) strengthen due
diligence and liability incentives; and (v) create demand Side incentives, including
facilitating product differentiation strategies and provide stronger economic signals to
producers.

Many of the roles of traceability are served by the functions in which the systems
serve. Traceability serves as a reactive function, which means reduction of both the cost
of a recall and damage to a reputation caused by a delayed or slow recall. Traceability
also allows liability for food/product safety problems to be more easily established along
the supply chain, hence, there is an incentive for firms to produce safe, high quality, food.
Furthermore, information costs arising from quality verification for consumers are
reduced by facilitating the labeling of credence attributes, including those related to food
safety, animal welfare, environmentally-friendly production practices, etc. (Golan et al.,
2003; Hobbs, 2003). Two final functions that these traceability systems perform is to

provide information ex ante on quality attributes, enabling consumers with an ethical

objection or food safety concerns to avoid the product and the systems may help prevent
or punish false labeling (Hobbs, 2004).

Hobbs (2004) notes that the key features of a traceability system depend on the
attribute that drives its development and highlight the need to consider fully the nature of
the information asymmetry problem before implementing a traceability system. The
diversification in the roles and functions of traceability systems contribute greatly to how
and why traceability systems differ. Further discussion and analysis in this thesis will be

based on live animal traceability, leaving meat traceability issues for future research.

2.1.2 Live Animal Traceability

Live animal traceability can be accomplished via a variety of systems; an example
being NAIS which is currently scheduled to be fully implemented in 2009 (Bulut and
Lawrence, 2007). The main function of these live animal traceability systems is to
quickly identify agricultural premises exposed to an animal disease so that the disease
can be more effectively controlled or eradicated. The USDA (2005) has set a long-term
goal of 48 hour traceback. Furthermore, live animal traceability provides proactive
information and quality verification, which is essentially an increase in production
information available to producers throughout the supply chain.

More specifically related to livestock and in this case beef cattle, traceability is a
system that can identify individual animals or groups of animals, the premises where they
are located, and the date of entry to each premises. Traceability systems vary greatly
with some systems being deep and tracking beef and beef cuts from the retailer back to

the farm, while other systems only extend back to a key point in the production process.

10

Some systems are very precise tracking product to the minute of production, exact
premises produced at, and animals commingled with. Others are less precise, tracking
cattle only to premises in a large geographical area. Finally, some traceability systems
collect and track information on a broad range of attributes, while others collect and track
only a few.

There are certain differences among traceability systems and in order to
understand these differences, standard characterization of these systems is helpful and
assists in defining the function of a particular system. These standard characterizations
may be made in terms of their breadth, depth, and precision (Golan et al., 2004). Breadth
is the amount of information recorded by the traceability system and could include
information such as age verification, production practice information,
performance/ genetic information, health records as well as a host of additional
information. Depth describes how far back or forwards the system tracks. A deeper
system will enable the establishment of links among more agents further up or down the
supply chain, where as a broader traceability system enables tracking of a larger variety
of attributes throughout supply chains. In most cases, the depth of a system is largely
determined by its breadth because once the firm or regulator has decided which attributes
are worth tracking, the depth of the system is fundamentally determined (Mus, 2006).
Precision reflects the ability of the system to pinpoint an attribute of interest. One
example in the supply chain may be pinpointing a particular wholesale or retail cut of
beef to an animal or lot of beef to a processing plant. That is, the unit of analysis used in

the system and the acceptable error rate determines precision (Golan et al., 2004).

ll

Ever since traceability has been brought to the public eye, the concept has been
confused with that Of animal ID. Animal ID is indispensable for live animal/meat
traceability to work effectively, but in addition to the ID of animals, traceability often
requires premises to be identified and cross-checking of data. Animal ID is critical for
proving ownership and providing a means to track or trace animals; so, coupled with
premise registration and data management, ID can be utilized in a traceability system. It
is important to note that ID and traceability are not ends in themselves, but rather means
or tools available to achieve a given objective. To the extent that it is understood that
traceability is a tool, that ID is one of the aspects of traceability, and that it goes hand-in-
hand with the analysis of recorded information, the various components involved can be
successfully implemented (Ammendrup and Barcos, 2006). Traceability systems allow
producers and animal health officials to respond more quickly and effectively to animal
disease outbreaks, provide the basis for certifications, and provide valuable production
information for producers and consumers.

Because no single recipe can be provided for traceability, responsibility and
design typically falls on the different sectors, trading partners, and parties defining the
objectives and implementing the traceability systems. Undeniably, economic and
technical decisions on which type of traceability system should be designed and
implemented involve trade-offs between system features and their related benefits and

costs (Souza-Monteiro and Caswell, 2004).

12

2.2 Economics and Current Status of Traceability Systems

2.2.1 Economics of Traceability Systems

Economics of traceability systems involves describing the economic incentives
motivating traceability systems. The economic incentives pushing these new systems
originate from the forces changing the meat marketplace and include improving animal
health management and rapid response systems, meeting consumer demands for meat
safety, maintaining and building international trade, verifying product credence attributes,
properly assigning liability, and in improving management throughout the meat supply
chain (Tonsor and Schroeder, 2006).

Traceability is key to improving animal health management because it allows for
quick and effective response to an animal disease event (whether it is a single incident or
a full-scale outbreak). Retrieving animal location and movement data within a short time
frame (i.e., USDA (2005) long-term goal of 48 hour traceback) is necessary for efficient,
effective disease containment.

Consumers have become increasingly concerned about the processes (i.e., inputs
and methods) used to produce the beef that they eat and the intrinsic quality attributes
(i.e., tenderness and nutrition) that beef possesses. Food safety and food quality issues
have moved to the forefront of consumer concerns; this is much attributed to high profile
food safety scares.

If competitors are able to differentiate their beef and beef products as being
superior to US. products in terms of the additional attributes provided and verified, the
US. may lose market share in various international export markets. A prime example of

this is the effect of food safety concerns on Japanese markets, including the discovery of

13

BSE in the US, which lead to heightened import restrictions and regulations. Japan was
(from 1991 to 2003) the U.S.’s principal export market for beef and such concerns have
lead to a loss of US. market share, because competitors such as Australia have succeeded
in convincing Japanese buyers that their products are “safer” than US. products because
their system provides more assurances than the US. system (Dickinson and Bailey,
2002). Finally, consumers may simply be willing to pay for traceable beef and a market
opportunity may be lost to US. producers if such products are not produced in the US.
(Dickinson and Bailey, 2002). Therefore, the increasing likelihood of losing export
markets, due to the failure to instill confidence in foreign consumers (or in foreign
political leaders) of the beef industry’s ability to produce safe food, offers an increasing
return to implementing a traceability system in the US. beef industry (Golan et al.,
2004).

Traceability provides information on the quality of the product, because
traceability systems allow producers and consumers to observe more of the production
process. Traceability systems may help to verify the existence of credence attributes and
can instill additional confidence in consumers that they are in fact purchasing a product
possessing the characteristics they desire (Tonsor and Schroeder, 2006). Credence
attributes are those quality attributes that are of concern for the consumer, but where no
clues are accessible in the process of buying and consuming to confirm the attributes
existence (Becker, 2000; Tonsor and Schroeder, 2006). Some examples of credence
attributes offered by Hobbs (2003c) include: enhanced food safety practices on the farm

or in the processing plant. Alternatively, they may identify credence attributes with

14

respect to the reduction of environmental extemalities or those related to ethical
preferences with respect to animal welfare.

Tonsor and Schroeder (2006) indicate that traceability systems do not alter the
liability of an event; however, they can provide useful information in properly accessing
legal responsibility by those involved in the production chain. An economic argument
for adopting traceability systems arises from the threat of civil legal action against a firm
producing unsafe food and the resulting financial damages including legal penalties,
damages to a firm’s reputation, and its loss of brand capital. The ability to trace products
allows liability for food safety problems to be more easily established along the supply
chain and reduces the monitoring and enforcement costs for consumers and downstream
food distributors by identifying the party at fault and in seeking legal compensation
(Hobbs, 2004)

Golan et al. (2004) and Mus (2006) defined improving management throughout
the beef supply chain as the ability of a firm to reduce the cost such as movement,
storage, and control of products in the supply chain and listed this as a main determinant
for a company to be successful. Companies operate in the food industry where profit
margins are very low, thus, supply-side management has become increasingly important
for firms to remain competitive in the market. Therefore, an effective and efficient
traceability system is a key factor to reducing the cost associated with the above given
supply-related activities. Brester (2002) and Tonsor and Schroeder (2006) also note that
implementation of a traceability system in the beef industry may aid in bringing the beef
industry’s ability to transfer information throughout the production process and become

more competitive with the pork and poultry industries.

15

2.2.2 Current Status of Traceability Systems

Golan et al. (2004) report the existence of several beef traceability systems in the
US. Though state authorities have promoted some, current systems have been mainly
private and market driven. The National Animal Identification System is the well known
federal beef traceability system which evolved from previous efforts to implement a
national voluntary beef traceability system. The National Institute for Animal
Agriculture assembled a task force in April 2002 to create the NIWP, which was later
called the USAIP as presented in 2003. The USAIP called for the establishment of
individual premise ID by the summer of 2004, individual animal ID by 2005, and full
implementation and compliance (all covered Species and their movements - both
interstate and intrastate) by July 2006 (Bailey and Slade, 2004). When USDA adopted
the plan it was renamed NAIS. The National Animal Identification System has become
quite well known among US. livestock producer groups after December 2003 when a
dairy cow in the state of Washington was diagnosed with bovine Spongiform
Encephalopathy (BSE or Mad Cow Disease). Many of the efforts and goals of NAIS, as
it was originally developed and implemented, saw opposition and encountered many
obstacles. The National Animal Identification System was initiated to enhance
previously existing disease programs through the establishment of standards that could be
used for all state/federal disease programs nationwide. The focus of NAIS is on animals
that enter commerce, that is, those animals that move from their farm and ranch to
markets and/or locations where they commingle with animals from other premises. This
is where the impact of a disease is the greatest, both in terms of value of animals and the

potential cost of lost production (Cattle Network, 2008).

16

The following NAIS component descriptions were taken from the NAIS website
(USDA APHIS, 2008). Premise registration is the foundation ofNAIS and is seen as
fundamental to containing animal diseases. Owners of premises involved in production
or commerce of animals can voluntarily register their premises with their state or tribal
animal health authority. Once registered, their premises are assigned a unique premises
identification number (PIN) that corresponds to the contact information that was
voluntarily provided. Producers who choose to participate in voluntary premises
registration will be notified quickly when a disease outbreak or other animal health event
might put their animals at risk. It is important to note that registering premises does not
require producers to participate in the other two NAIS components (animal ID and
animal tracing as described below). As of July 6, 2008 only about 32.86% of the
1,438,280 estimated premises nation-wide were registered (USDA APHIS, 2008).

Animal ID is the second voluntary component of NAIS. Animal ID, whether
individual or group/lot provides producers and owners a uniform numbering system for
their animals to help manage them more closely. The individual animal identification
number (AIN) is unique and stays with the animal for its lifetime. This number allows
the data base to link the animal to its birthplace or premises of origin; when combined
with animal tracing, the AIN also allows the data base to link the animal to each
premises/location that has been reported for it. The Animal and Plant Health Inspection
Service allow firms to use supplemental technologies such as a RFID and compatible ear
tags as a part of their identification system (USDA APHIS, 2008; Mus, 2006). Animal
ID offers a valuable tool for producers and owners whose animals enter commercial

production or move to locations where they come into contact with animals from

17

multiple/other premises. In these situations, there is an increased potential for the
animals to be exposed to or impact the spread of disease. Individual ID is many times the
standard for animal ID; however, group/lot identification is best suited for animals that
"stay together" and are raised as one group (e.g., poultry). When animals "stay together,"
individual ID of each animal in the group is not necessary because it would not enhance
disease response efforts. In addition to protecting animal health, animal ID is many times
used as a valuable tool for other, "non-NAIS" purposes - such as animal management,
genetic improvement, and marketing Opportunities. When used in conjunction with other
NAIS components, animal ID can also help protect producers’ access to markets. If a
disease outbreak or other animal health event occurs, and a producer’s animals are not
linked to any affected premises or areas, they could use animal ID numbers and
movement records (included in the third NAIS component) to demonstrate that their
animals are disease free. Producers may choose whether to submit their information to a
privately-held or state-held database. Animal health officials will only request access to
animal ID records in the case of an animal health event (USDA APHIS, 2008).

Animal tracing is the final component of NAIS and is under development by
states and the private sector. Once this component is complete, animal tracing should
offer an additional option to improve animal management and better protect animal
health. Producers will be able to choose an animal tracking database (owned and
operated by private industry groups or states) and report certain animal movements that
might pose a significant risk of disease transmission. When there is a disease outbreak or
other animal health event, the animal tracking databases provide timely, accurate records

that Show where animals have been and what other animals have come into contact with

18

them. Once the animal tracing component is complete, there will be several important
points to consider when choosing whether to participate. Participating in animal tracing
helps animal health officials receive accurate information about where a disease outbreak
or other animal health event is occurring. Under NAIS, USDA will not have direct
access to animal movement records. Private or state databases will house and maintain
information regarding animal movements. Federal and state animal health officials will
request access to this information only if a disease or animal health event - such as an
outbreak of avian influenza or brucellosis - occurs. Federal law protects individuals'
private information and confidential business information from disclosure (USDA
APHIS, 2008).

Private sector traceability initiatives in the beef industry include individual supply
chain initiatives and industry-wide programs. Within some of these programs, firms
provide voluntary labeling of credence attributes and sometimes these programs are
supplemented by third party certification. Credence attributes are defined as attributes
that cannot be determined even after purchase or consumption, such as animal welfare or
organic production (Weiss, 1995; Roberts et a1 ., 1996). The credence nature of food
safety and quality attributes may lead to markets being dominated by low-quality
products if producers of high-quality (or “safer”) food are unable to offer credible
assurances to consumers (Golan et al., 2003; Hobbs, 2004). Supply chain partnerships
delivering traceability have emerged for multiple reasons including to help deal with the
loss of consumer confidence. The meat processing sector has also recognized the
potential role of traceability in bolstering consumer confidence, and as a product

differentiation strategy (Hobbs, 2003b). The emergence of traceability systems in the

19

private sector can also be seen as a result of pressure from downstream food retailers who
are motivated by the desire to reduce risk exposure or reduce the information costs of
monitoring product quality or downstream production methods (Hobbs, 2003b). Finally,
industry associations or producer groups have been responsible for introducing industry-
wide private sector traceability programs.

Many of these US. private industry associations or producer group traceability
systems tend to be motivated by economic incentives, not government traceability
regulation. These private systems allow for the verification of many USDA accredited
claims, such as age and source verification, organic, natural, etc. On the private, state,
and national level the USDA has utilized Process Verified Programs (PVP) and Quality
Assessment Programs (QSA) that have been historically used for verification purposes in
many industries and for a variety of products. In the case of US. beef cattle the USDA
has established PVP’s and QSA’S to ensure the credibility and authenticity of the process
claims being made about traceable beef products. This includes claims such as: age and
source verified, organic, etc. Currently PVP and QSA programs exist to back these
claims and traceability systems are implicit in PVP and QSA programs for ensuring
credibility and authenticity. Some countries require US. beef exporters to be accredited

under a USDA Export Verification Program.

2.3 Consumer Impacts/Preferences
Past studies have focused on the value of information that characterizes products
that could either be placed on labels or communicated to consumers in other ways. For

example, research has recently focused on consumer acceptance of and government

20

policy toward genetically modified (GM) food products (e. g., Caswell, 2000; Huffman et
al., 2003a,b; and Lusk, Roosen, and Fox, 2003). Other research has examined the value
to consumers of providing information of different single or bundled characteristics,
including certifying enhanced food safety, the processes used to produce food, the
location in which food was produced, or the certifying agency (e. g., Dickinson, Hobbs,
and Bailey, 2003; Loureiro and Umberger, 2006). A few studies have addressed the issue
of traceability directly and have found traceability to be a valuable characteristic in food
products (e. g., Dickinson and Bailey, 2002).

Dickinson, Hobbs, and Bailey (2003) examined consumer WTP in Canada and
the US. for traceability, transparency, and enhanced quality assurances (TTA)
characteristics in red meat products. According to Liddell and Bailey (2001) traceability
is sometimes called identity preservation and is defined as the ability to track the inputs
used to produce food products backward and forward to/from their source at different
levels of the marketing chain. Baines and Davies (1998) and Early (1998) indicate that
transparency refers to the public availability of information on all the rules, procedures,
and practices used to produce a food product at each level of the marketing chain.
Enhanced quality assurances that can be provided by TTA are referred to as “extrinsic”
qualities by Baines and Davies (1998) and are characteristics that affect neither food
safety nor typical government grading, but which are still valued by consumers. This was
an important analysis because valuable comparisons were identified considering the
Canadian red-meat industry was moving toward more TTA, especially traceability, while
the US. red-meat industry had been much slower in adopting TTA protocols. A sealed-

bid Vickery style auction was the main instrument used to gain information and followed

21

the basic design suggested in Shogren et al. (1994) for eliciting bids to upgrade a meat
sandwich, thus determining consumers WTP for traceability and assured quality
assurances. Results for both countries were very similar in how preferences for the
characteristics were ordered, and they were also similar in comparing the average bids for
characteristics, hence a very close correlation was found between consumers’ WTP in
both countries. One important note from this study was that traceability, while receiving
positive bids, was the least valued of the three individual characteristics presented to
consumers (e. g., animal welfare, food safety, and traceability). Dickinson, Hobbs, and
Bailey (2003) suggested that traceability should be bundled with other characteristics that
could be verified with traceability when food products were marketed with these
characteristics.

Another related study by Loureiro and Umberger (2006) determined the relative
value US. consumers place on several beef attributes including: traceability, country-of-
origin labeling (COOL), food safety inspection, and tenderness. A mail survey was used
to solicit information regarding respondents’ purchasing behavior and attitudes about
beef products, beef qualities found most desirable, food safety attitudes, questions
involving a choice modeling experiment, and socio-demographics. Findings in this study
were comparable to that of Dickinson, Hobbs, and Bailey (2003) which found that
consumers placed the highest relative value on food safety certification.

In a study addressing the issue of traceability directly results by Dickinson and
Bailey (2002) suggest that although traceability for beef products was found to be valued

to some extent; subjects placed an even larger value on specific attributes that might be

22

verified by a traceable meat system. Bids for beef traceability were statistically lower

than bids for both animal treatment assurances and bids for increased food safety.

2.4 Work Needed

The only known economic study examining beef industry perceptions and
preference is Bailey and Slade (2004) who conducted a survey to measure the level of
support among state veterinarians and representatives of producer groups. They
examined how support for a Specific animal ID proposal (USAIP) varied based on
concerns about animal health and the perceived costs and benefits accrued to different
levels of the marketing chain. Of most importance are the representatives of the producer
groups because producers will likely incur a majority of the costs. The study found that
over 90% of state cattle producer association respondents indicated support for a national
cattle ID program, while only 41% indicated that they supported the USAIP (Bailey and
Slade, 2004). The results of this study provide evidence Showing that producers do
support traceability. The lack of confidence in initial programs, as indicated by the 41%
in favor of the USAIP, documents the need for determining the most important
characteristics/attributes of voluntary traceability systems to aid in design and promotion
of a more accepted traceability system. This in turn should help increase voluntary
participation rate amongst producers in voluntary systems.

A majority of the past research has sought to analyze consumer’s perceptions
towards voluntary traceability systems; however, there has been little research on
producer’s perceptions towards these same traceability systems. This thesis sets forth

models that will examine cow-calf producer’s preferences for traceability systems and

23

system attributes which may be characterized by heterogeneity. Throughout this thesis,
the term preference heterogeneity refers to variability or differences between preferences
of producers. Accounting for heterogeneity of producers will be usefiil in estimating
unbiased models. Incorporating and understanding heterogeneity may provide

information on the distributional effects of traceability policy alternatives.

24

CHAPTER 3: OBJECTIVES

The overall purpose of this research was to identify U.S. cow-calf producer
preferences and perceptions regarding voluntary beef traceability systems. Meeting this
core objective allowed for better identification of the potential success of alternative
voluntary traceability systems that could exist in the beef industry. This “potential
success” was primarily measured by producers’ preferences for traceability systems
varying in attributes including premiums and discounts, entities maintaining traceability
data, and the quantity and type of maintained information. This analysis sought to build
upon the existing literature and prior traceability system studies by gathering and
analyzing survey data from cow-calf producers to allow for the economic analysis of
various voluntary traceability systems. Survey data was used to parameterize the
economic analysis and inform the discussion regarding implications of traceability
system design and promotion.

This research sets the groundwork for identifying participation rates for various
traceability programs, which future research could utilize to obtain conclusions regarding
animal disease response implications. Increasing voluntary participation rate Should
allow producers and animal health officials to respond more quickly and effectively to
animal disease outbreaks in the US. Furthermore, a national livestock identification
system can be used to launch more extensive quality assurance programs. This type of
traceability system may be used as a platform on which additional quality assurances can
be provided to producers further down the supply chain and to consumers. Traceability
systems that are most aligned with the preferences of cow-calf producers will experience

higher voluntary participation. However, traceability systems based solely on cow-calf

25

producer preferences may not maximize the nation’s ability to respond to animal disease
or meet alternative goals of nationwide traceability systems. Producers and animal health
officials must be conscious that lower voluntary participation in a stringent system may
well be better than higher voluntary participation in a weaker system for accomplishing
many of the traceability system initiatives and goals.

In order to achieve the main purpose of the study as laid out above, the research
was set up where a first tier of objectives was identified as well as a second tier of
objectives. These objectives identified hypotheses to be evaluated which provided
guidance in subsequent chapters in properly analyzing these objectives. An outline of the
tier 1 objectives is as follows:

0 Tier 1: Determine preferences of cow-calf producers for alternative voluntary
traceability systems and system attributes.

0 Objective 1.1: Evaluate how sensitive producer preferences are to price
adjustments (premiums/discounts).

0 Objective 1.2: Examine if producer preferences are sensitive to the entity
in charge of data maintenance.

0 Objective 1.3: Examine how the inclusion of additional information
requirements affects voluntary participation rate.

0 Objective 1.4: Investigate how producers’ welfare is affected if
alternative levels of traceability become mandatory.

0 Objective 1.5: Identify if producer preferences are sensitive to producer

demographics, perceptions, and current production/technology practices.

26

The core objective of this study was to determine what preferences cow-calf
producers have for alternative voluntary traceability systems and system attributes. This
core objective led to evaluating a key hypothesis of: “cow-calf producer preferences for
voluntary traceability systems are homogeneous.” This hypothesis was evaluated using a
choice experiment designed to accomplish two main things: (1) identify “average” or
representative preferences and (2) examine the extent of heterogeneity in these
preferences. According to Ouma et al. (2007) the conceptual framework for choice
experiments arises from the consumer theory developed by Lancaster (1966), which
postulates that preferences for goods are a function of the traits or characteristics
possessed by the good, rather than the good itself. Therefore, analyzing the reasoning for
an observed choice can be done by examining the attributes/components of the chosen
and not chosen alternatives. In turn, a value or preference for these
attributes/components can be derived.

Another objective was to evaluate how sensitive producer preferences were to
price adjustments (premiums/discounts). This led to a set of evaluations to examine if
producers are price sensitive and if this price sensitivity varied across producers. This
was accomplished by evaluating heterogeneity in producer preferences for premiums and
discounts to examine if all producers are equally price-sensitive.

Another objective was to examine if producer preferences were sensitive to the
entity in charge of data maintenance. Thus, answering the questions, do preferences for
particular systems change if the entity maintaining the data switches from government to
private and do preferences vary if the private entity is based within the cattle industry or

not? An additional objective stemming from the choice experiment was to examine how

27

the inclusion and composition of additional information (e. g., age verification vs.
production practice information) of Advanced Traceability systems affected voluntary
participation rate.

Continuing under the same umbrella of the core objective of this research,
additional objectives were identified followed by hypotheses to test. An objective was to
investigate how producers’ welfare was affected if a certain level of traceability becomes
mandatory (e. g., removal of N0 T raceabilily option). This may arise as mandatory
traceability systems could be introduced to correct perceived market failures when firms
fail to supply the socially optimal level of traceability (Hobbs et al., 2005). Market
failure can occur because the credence nature of food safety and quality attributes may
lead to markets being dominated by low-quality products if producers of high quality
food are unable to Offer credible assurances to consumers (Golan et al., 2003; Hobbs,
2004; Hobbs et al., 2005). Alternatively, traceability systems facilitate the traceback of
products in the event of a food safety problem, reducing the impact on public health and
protecting the reputation of all firms in the same industry, thus, net social benefits of a
traceback system may outweigh the net private benefits, leading to underinvestment in
traceability (Hobbs, 2003; Golan et al., 2003; Hobbs et al., 2005). Initially, the USDA
stated that NAIS would start as a voluntary program and later become mandatory to
achieve full participation with premises registration and animal ID to be required by
January 2008 and the reporting of defined animal movements to be required by January
2009. However, in late 2006, the agency decided that NAIS would remain voluntary
(United States Government Accountability Office, 2007). This leads to analyzing the

cow—calf prOducer welfare impacts of having voluntary traceability as an option as

28

opposed to mandatory traceability as the only option where the N0 Traceability option is
prohibited. We also estimated the welfare effect producers would experience given the
removal of both No Traceability and Advanced Traceability alternatives. This analysis
will be beneficial as the US. Government Accountability Office (GAO) report indicated
that industry association officials suggested that if NAIS became mandatory, producers
who have voluntarily participated would lose the market advantage they currently enjoy
through higher prices paid at market or Slaughter for animals they identify for marketing
or management purposes.

A final hypothesis under this core objective that was identified was: “producer
preferences are sensitive to producer demographics, perceptions, and current
production/technology practices.” Then, if it was determined that producer preferences
were sensitive to producer demographics, perceptions, and current production/technology
practices it was determined how their choice of traceability systems and traceability
system attributes was affected.

Overall, the evaluations listed above drive the objective of designing these
voluntary traceability systems in order to maximize expected voluntary participation rate,
dually, showing how these voluntary traceability programs should look given alternative
goals for traceability systems. Overall, immediate guidance in how voluntary traceability
programs should be designed stemmed from our choice experiment and associated
models.

A second tier of objectives allowed for an alternative evaluation of individual

perceptions of traceability systems. An outline of the tier 2 objectives is as follows:

29

Tier 2: Evaluation of individual perceptions of traceability systems.

0

Objective 2.]: Identify cow-calf producer forecasts of voluntary
participation in NAIS with the registration of premises and registration of
premises with the use of RF ID.

Objective 2.2: Determine what type of producer and/or operations have
premises registered in NAIS.

Objective 2.3: Determine what type of producer is in favor of making
NAIS mandatory.

Objective 2.4: Determine if certain concerns and issues were still of top
apprehension to traceability system participation or have producers
perceptions shifted within the industry.

Objective 2.5: Identify if producer preferences are sensitive to producer

demographics, perceptions, and current production/technology practices.

Continuing with this second tier of objectives, two sets of producer’ forecasts
were then identified to acquire even more valuable information regarding U.S. cow-calf
producers. Obtaining forecasts of voluntary participation in NAIS with the registration of
premises and registration of premises with the use of approved animal identification
devices (i.e., RFID) on cattle leaving the premises met the forecasting objective. By.
utilizing these forecasts, and the characteristics that describe the most optimistic and
pessimistic forecasters, a better sense of how traceability systems should be designed and
promoted in order to maximize voluntary participation rate was identified.

An evaluation of what type of cow-calf producer has their premises registered in

NAIS was performed. Furthermore, what type of cow-calf operator is in‘favor of making

30

NAIS mandatory was determined. These two objectives were investigated judiciously
because it was important to determine if producers are in favor/not in favor of mandatory
NAIS because their premises were already registered in NAIS, or alternatively, did their
characteristics and/or perceptions lead to their decision concerning mandatory NAIS.

Previous studies (e. g., Kansas State University, 2006) identified cost to producer,
reliability of technology, confidentiality of information, and liability to the producer as
top concerns of cow-calf producers. Given this previous research, an objective of this
work was to determine if these concerns, as well as additional concerns and issues, were
still of top apprehension to participation, or have producers perceptions shifted within the
industry. Producers reaffirming these factors or changes in attitudes will most surely
affect the optimal design and voluntary participation rate. Once again, the typology of
producers through the hypothesis, “concerns are sensitive to producer demographics,
perceptions, and current production/technology practices,” was revisited in order to
determine if their previous concerns held certain.

Chapter five, Conceptual Framework and Models, introduces random utility
theory which underlies objectives related to choice experiments (e. g., Tier 1 objectives)
and was the basis for meeting the core objective. Also, unfolded within the conceptual
framework chapter is theory based around the forecasting objectives. Chapter 6,
Empirical Modeling, shows how each of the objectives was met econometrically utilizing

Tobit, Probit, MNL, RPL, and LCM models.

31

CHAPTER 4: SURVEY AND DATA DESCRIPTION

4.1 Survey Design

A survey was designed to obtain information from US. cow-calf producers
regarding demographics, production practices, and potential beef traceability systems (the
original survey is provided in Appendix 1). The survey was conducted by Michigan
State University faculty and graduate students in conjunction with BEEF Magazine who
supplied the mailing list for the survey. The random selection of farms to receive the
survey allowed equal opportunity for selection regardless of participation in various farm
organizations; however, given that BEEF Magazine subscribers traditionally have herd
sizes greater than one-hundred animals, the sample was not expected to be completely
representative of the diverse population of US. cow-calf operations. Thus, conclusions
are drawn only for the producers surveyed.

The comprehensive survey included questions regarding various aspects of cow-
calf production, including demographics and current production/technology practices,
perceptions concerning traceability, and a choice experiment focused on beef traceability.
The survey data collected was used to parameterize the analysis of cow-calf producer
characteristics, perceptions, and choices affecting the implementation of individual
animal traceability systems.

Questions regarding gender, producer’s age, state of residence (U.S. farm
production region), farm organization(s) of membership, educational description, years
raising beef cattle, expected years raising beef cattle, estimated annual pre-tax income,
household income from off-farm sources, operation’s labor supplied by non-family (paid

employees), operation's feed/forage needs produced on farm, marketing claims,

32

marketing methods, number beef cows that calved in 2007, operation’s sales at particular
production stages, and animal identification methods currently used were asked to better
understand the characteristics of the cow-calf producers and their operations.

More in-depth questions concerning cow-calf producers’ perceptions of important
issues and concerns to the US. beef industry when designing a national, individual
animal traceability system were then asked to capture the most important issues and
concerns of cow-calf producers. Furthermore, respondents were allowed to indicate their
level of agreement with statements involving economies of scale, liability, COOL, and
mandatory traceability when implementing individual animal traceability systems.
Appendix 2 provides a complete list of the variables elicited from these questions and
complete definitions of each as they will be used throughout subsequent discussion of
model specification, results, and conclusions.

A choice experiment was utilized to Simulate real-life situations in which cow-
calf producers choose between alternative traceability systems. Choice experiments
permit multiple attributes to be evaluated, thereby allowing researchers to estimate trade-
offs between different alternatives (Lusk, Roosen, and Fox, 2003). A reference page (see
Appendix 1) describing NAIS Traceability, Advanced Traceability, N0 Traceability, the
entities maintaining the data, the premium/discount (per animal sold), and additional
advanced traceability information was included before the choice experiment scenarios
for reference in interpreting the alternative traceability options.

In this choice experiment, cow-calf producers were presented with a set of four
scenarios, each of which involved choosing a preferred alternative fiom three different

traceability systems. The three traceability systems included: (a) NAIS Traceability, (b)

33

Advanced Traceability, and (c) No Traceability. NAIS Traceability, as the name
suggests, refers to NAIS and is a voluntary state-federal-industry partnership, which is a
modern, streamlined information system that helps producers and animal health officials
respond quickly and effectively to animal health events in the US. The NAIS program
consists of three components: (1) premises registration, (2) animal ID, and (3) animal
tracing. Premise registration requires a producer to provide basic contact information.
Animal ID (on an individual animal basis in cattle) provides the producer with a uniform
numbering system for identifying their animals and a way of linking those animals to
their birthplace or premises of origin. RFID ear tags was listed as the type of animal ID
used. Animal tracing allows the producer to choose an animal tracking database (owned
and operated by private industry groups or states) and report certain animal movements
that might pose a significant risk of disease transmission. The USDA is left responsible
for protecting individuals' private information from disclosure (USDA APHIS, 2008).
Advanced Traceability was considered in the choice experiment because
advanced systems besides NAIS are becoming ever more popular and available to
producers. These systems provide quality signals to consumers regarding experience or
credence attributes (Hobbs, 2004). Furthermore, advanced traceability systems reflect
demand-side incentives, including reducing information costs for consumers,
implementing product differentiation strategies, and providing accurate economic signal
to producers (Buhr, 2003; Meuwissen et al., 2003; Hobbs, 2004; Golan et al., 2004;
Smith et al., 2005). Finally, these advanced systems provide proactive information and
quality verification which is essentially an increase in production information available to

producers. Thus, it can be seen that advanced traceability systems depend on the

34

attribute(s) that drive their development. Advanced Traceability refers to a traceability
system with the same basic participation requirements as NAIS Traceability, but also
requires producers to record and provide additional information that is believed to be of
particular interest to beef consumers and/or is believed to improve production
management throughout the beef supply chain. This additional advanced traceability
information includes: age verification, production practice information,
performance/ genetic information, and health certifications/vaccinations records. Age
verification, performance/genetic information, and health certifications/vaccinations
records were only listed in the survey. Production practice information was described as
information that would include, but is not limited to, growth hormones used and/or grass-
fed diets used. The additional information requirements were not presented as individual
or group Specific, which allowed respondents to interpret these as they seen fit. More
generally, the additional information requirements were purposely not overly specific as
doing so would have limited the scope of this study (e.g., required valuation of fewer
attributes and/or levels). However, we do acknowledge that different perceptions of
producers in the requirements of these traceability system attributes likely impacted their
willingness to participate. As such, all of our conclusions are strictly based upon
producer responses to the information provided to them. Furthermore, participation in
this traceability system would require partaking in random verification audits to further
validate consistency between on-farm practices and information maintained within. the
traceability system.

As recommended by Adamowicz et al. (1998), a nO-choice option was also

presented to participants, because this is an obvious element of choice behavior. N0

35

Traceability refers to a scenario without participation in any individual animal
traceability system. This alternative was never associated with a premium and, therefore,
always presented with a discount greater or equal to $0.00 per animal sold. Accordingly,
managing entity and additional information to provide were absent from this choice
alternative.

Among the three choices for traceability systems, attributes were randomly varied
(following orthogonal fractional design procedures; Kuthfeld, Tobias, and Garratt, 1994)
in order to back out cow-calf producer’s preferences for traceability system attributes.
The attributes included were: (a) premium/discount per animal sold, (b) managing entity,
and (c) additional information. Options differed in terms of the premium or discount (per
animal sold) that a producer would receive by selecting each alternative. These price
adjustments ranged from discounts of up to $15 per animal (indicating you receive
SIS/animal less than the market price) to premiums of up to $15 per head (indicating the
producer would receive SIS/animal more than the market price). Within the survey,
negative numbers indicated discounts and positive numbers indicated premiums.

Alternatives for managing entities were included due to the growing concerns
among many producers regarding the collection and use of what they view as their
private production information (Becker, 2007). That is why some producers want a
private third-party, rather than USDA, to collect and maintain animal data. When
considering government (USDA) or private entity as the manager of traceability data,
producers are often concerned with who has the best qualifications for consistency of
data recording and management (including confidentiality assurances), ability to respond

to technical problems in the field, and the Speed of animal traceback. The entity

36

maintaining the tracing data and managing each traceability system took one of three
forms: Government, Private - Industry, and Private - Not Industry. A reference to
government means a government entity (such as USDA) manages and maintains the
traceability system. Private - Industry means a private entity manages and maintains the
traceability system. This entity is based within and owned by the beef industry.
Furthermore, this entity specializes in designing, managing, and maintaining traceability
Specifically for the beef industry. Private - Not Industry means a private entity manages
and maintains the traceability system. This entity is not based within, and is owned
outside of the beef industry. This entity specializes in designing, managing, and
maintaining livestock traceability systems.

The additional information required to provide in the Advanced Traceability
choice was presented in all possible combinations (e. g., 24 = 16) to ensure proper survey
design and exhaust all possibilities to allow for the identification of the most important
and most desired combinations of additional information to provide. This also allowed
for embedding of additional information combinations to be evaluated (e. g., does “WTA
for age verification + WTA for health records = WTA for age verification + health
records?”

A concern in most choice experiments is hypothetical bias. Typically, with
hypothetical questions, respondents will be more willing to choose to participate or not
participate in a voluntary traceability system than they would if real money and
circumstances were involved. A cheap talk script was included before the scenarios
which informed respondents of the hypothetical bias. This has been shown to reduce

hypothetical bias in choice experiment research (Lusk, 2003).

37

4.2 Survey Data

On November 26, 2007 a total of 2,000 (1,998 effective) surveys were mailed to
cow-calf producers (selected on an “nth” name basis by BEEF Magazine) throughout the
country. A one-dollar bill was included in the survey to potentially increase participation
and response (Gregory, 2008). The respondent pool was expected to include cow-calf
operations of greater than one-hundred animals due to the characteristics of BEEF
Magazine subscribers. Michigan respondents (5) were not included in the final data set
because of the mandatory nature of the state’s individual, beef traceability system.
Contrary to earlier expectations, 28.10% of producers indicated that their operation’s had
less than one-hundred cows that calved in 2007. The respondent pool provided 655
useable surveys (32.78% effective response rate). Consistent with Michigan State
University research requirements when administering a survey, respondents were
presented the option to decline to answer individual questions or portions of the survey at
their discretion, if they chose to participate at all. Furthermore, the protocol for this
survey and research was approved by Michigan State’s University Committee on
Research Involving Human Subjects (UCRIHS).l Summary statistics were computed for
all questions. Appendix 3 provides a summary of responses to the entire set of survey
questions. Throughout the presentation of the summary statistics, the “number reporting
or N Valid” accompanies each set of summary statistics, which indicates the total number
of usable responses to a given question. Many questions allowed a respondent to check
all answers which were applicable to the operation from a multiple choice list, and such
questions were analyzed by tabulating the total number of responses and computing

frequencies.

 

' Institutional Review Board (IRB); number X07-1014 approved on October 23, 2007.

38

Approximately 92% of the respondents were male, with the average age of the
sample being 58 years. Distribution of respondents across US. states and production
regions followed the National Agriculture Statistics Service (N ASS) numbers for cattle
operations (NASS, 2008). The USDA Economic Research Service (ERS) farm
production regions were subsequently adjusted to help eliminate over parameterization
problems in estimation. This was accomplished by combining the Mountain and Pacific
regions, Lake States and Northeast region, and Southeast region, Appalachia region, and
Delta States. These new adjusted farm production regions still maintained geographical

differences. Figure 4.1 provides a pictorial representation of these regions.

Figure 4.1. US. Regions

 

Northern
Crescent

   
    

I

Corn Belt

 

 

 

Appendix 2 provides a state by state categorization of these geographical regions.
Based on the 2007 calendar year, producers indicated that 60.91% of the operations had

less than two-hundred beef cows that calved. Plastic ear tags were the most commonly

39

used form of animal ID (87.89%), with RFID being the least used at 9.18%. F orty-four
percent indicated their operation’s premises were currently registered with NAIS.

Contained in Appendix 3 (Tables 23-3 0) is a summary of ordered responses
regarding cow-calf producer’s perceptions of the importance of certain issues and
concerns to the US. beef industry in designing a national, individual animal traceability
system. Furthermore, this summary provides cow-calf producer’s agreement with
statements concerning economies of scale, liability, COOL, and mandatory traceability
when implementing individual animal traceability systems. Also provided are results of
producers’ perceptions concerning the allocation of benefits and costs when
implementing traceability systems and forecasts for NAIS premise registration and RFID
use. If individual animal traceability systems were put in place, cow-calf producers
believe most of the benefits are distributed rather evenly, whereas they believe costs are
largely born by the cow-calf producers.

Fifty percent of respondents believe that NAIS should not be mandatory, 21%
indicated NAIS should be mandatory, and 29% of producers were undecided. Finally,
forecasts for US. cow-calf operations with premises registered in NAIS by December 31,
2008 averaged 40%, while forecasts for US. cow-calf operations with premises
registered in NAIS and RF ID used by December 31, 2008 averaged 31%.

The last section of Appendix 3 (Tables 3 1-3 7) provides summary tables of the
hypothetical choice experiment. Approximately 75% of the choice experiments were
returned completed. The remaining 25% were either left blank or partially completed.

These survey findings will be discussed in subsequent chapters with regards to

how they inform the underlying economic analysis of factors affecting which producers

4o

and/or operations currently participate or would choose to participate in certain
traceability systems. Economic and management implications will then be parameterized
using the results to better enable the designing of individual, voluntary beef traceability

systems.

41

CHAPTER 5: CONCEPTUAL FRAMEWORK AND MODELS

In this chapter, the conceptual framework and models corresponding to the
objectives outlined in chapter 3 and underlying the empirical models that will be
estimated are developed. This involves describing the economic theory driving these
model specifications. The model development discussion is followed by a discussion in
the following empirical modeling chapter concerning evaluation procedures used in
examining the appropriateness of the developed models and the techniques and variables

employed. To further aid in following this progression, Figure 5.1 is provided as a guide.

Figure 5.1. Conceptual Model Progression Flow Chart

 

 

 

 

/ Conceptual Framework

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5.1
Forecasting Framework
1
5.1.1 5.2
Tobit Model Probit Model Specifications 53
1 Choice Model Specifications
5.2.1 1
Multinomial Probit 5.3_1
[ Multinomial Logit
5.2.2 I
Binary Probit 5.3.2
1 Random Parameter Logit
5.2.3 T
Endogeneity Bias 5.3.3
I Latent Classification Model
5.2.4
Trinomial Probit
L
5.2.5
Ordered Probit

 

 

 

 

 

 

42

5.1 Forecasting Framework

When analyzing producer intentions, an important question is what factors cause
intention development. The presumption is that beliefs are key elements in forming
attitudes, intentions, and eventually influencing behavior (Han and Harrison, 2006).
Beliefs represent the base set of information a producer has about an object or concept
(Fishbein and Ajzen, 1975). Thus, these beliefs will describe all the thoughts a producer
has about systems in association with various attributes, and beliefs play an important
role in forming attitudes towards action.

When producers are asked to forecast participation and usage rates (e.g., NAIS
registration and RFID use) concerning aspects of voluntary traceability systems their
beliefs about the future may be based on their current attitudes and characteristics. Thus,
providing insight into the characteristics of the most optimistic and pessimistic
forecasters. Given that forecasts by design range from 0 to 100% (Appendix 1, survey
question 24); to analyze producer stated forecasts of participation and usage rates, Tobit
models were utilized where forecasts were modeled as zero for a fraction of the
population but are roughly continuously distributed over positive values (Wooldridge,

2003).

5.1.1 Tobit Model

The Tobit model developed by Tobin (195 8) supposes that the decision to
participate in the market is- the same as the decision about the quantity or extent of
participation. This implies that any variable that increases the probability Of nonzero

value must also increase the conditional mean of the positive values.

43

Compared with the Probit model as described in a later section, which is based on
the cumulative distribution and estimates the probability of the dependent variable lying
inside at a 0-1 interval, the Tobit model was adopted, as it does not throw away any
information on the value of the dependent variable (Gong et al., 2007). In situations
when there is an upper or lower bound on an outcome variable, an appropriate statistical
model to apply is a Tobit model with left (lower) and/or right (upper) censoring.

The censored regression or Tobit model is appropriate when the dependent
variable is censored at some upper or lower bound as an artifact of how the data are
collected or measured (Tobin, 195 8). The general formulation that is typically presented
is for censoring at a lower bound of zero and is usually given in terms of an index
function (Greene, 2000):

VI =XiB+€i €~N(0,02)

(5.1) yi=0 ifyfso,
Yi=YI ifYI>0.

where the index variable, sometimes called the latent variable, E[y: | Xi] is x'iB , s is the
error, and B is a vector of coefficients to be estimated. For the ith observation, y: is an
unobserved latent continuous variable, Yi is the observed variable, x; is a vector of
values on the independent variable or explanatory variables, and it is assumed that 8: is

uncorrelated with x; and is independently and identically distributed. This specification

can also be written to encompass the more general double bounded range in which:

44

y: =x§B+ai 8~N(0,0’2)
y: =YI ifL<yj <U.

(5.2) *

Y: = L If yi S L

yi = U if y: 2 U

where L and U take on the values of the lower bound and upper bound, respectively.
The Tobit model has some important features that make it appealing for use with

our data. In addition, the Tobit model is consistent and unbiased provided that there is no

heteroscedasticity in the error terms.

5.2 Probit Model Specifications

A common behavioral assumption underlying economic theory and applied
research in economics states that “agents” aim to maximize their expected utility. This
research and analysis also relied upon these assumptions. The following specifications of
random utility models allows for the elicitation of preferences for complex
multidimensional systems, from which models of producer preferences may be estimated

(Hall et al., 2004).

5.2.1 Multinomial Probit Model

Following Greene (2000) unordered-choice models can be motivated by a random
utility model. Thus, for the ith individual faced with j choices, suppose that the utility of
choice j is (Greene, 2000):

(5.3) Uij = Z'UB + sij'

45

Then if the individual makes choice j in particular, then we assume that Uij is the

maximum among the j utilities. Hence, the statistical model is driven by the probability

that choice j is made, which is (Greene, 2000):

(5.4) PI'Ob(Uij>Uik) Vkij.

Probit models can be used to model any discrete choice/selection Situation. The
random utility framework was first utilized in Probit model specifications in analyzing
producer behavior in discrete choice situations. The multinomial Probit model, which
assumes that decision makers may be modeled as coming from a population of random
utility maximizers, where the error component is in the (unobserved) utilities arise from a
multivariate normal distribution (McFadden, 1981; Bunch and Kitamura, 1991). Now,
the probability that individual It selects alternative j is given by:

(5.5) Prob[UnJ- > Uni] v i .4 j

where this equation has a utility function in which alternative j is only chosen if it yields

the highest utility across individuals n.

5.2.2 Binary Probit

In our application of binary Probit models it is assumed that producers attempt to
maximize their utility when they face a binary choice. An example application in this
work is the ‘yes’ or ‘no’ decision of registering premises within the NAIS system (e.g.,
question 22 of the survey in Appendix 1).

Following Wooldridge (2003) in a binary response model, interest lies primarily
in the response probability:

(5.6) Prob[Y =1 |x]= Prob[Y =1 |x1,x2,...,xk],

46

For specifying the Probit models consider a class of binary response models of the form:
(5.7) Prob[Y =1 |x]= G[B0 + lel + + kak] = G[B0 + x13] ,
where G is a function taking on values strictly between zero and one: 0 < G(z) < l, for all

real numbers 2. This ensures that the estimated response probabilities are strictly between
zero and one. This estimation and subsequent estimation of Probit choice models will be
based on the method of maximum likelihood. The nonlinear Probit model for the
function G ensures that the probabilities are between zero and one. G is a standard

normal cumulative distribution function (CDF), which is expressed as an integral:

(5.8) G(z)=¢<z)a NOW.

—00
where 6(2) is a standard normal density:

(5.9) ¢(z) = (210‘1 ’ 2 exp(—zz /2).

Wooldridge (2003) shows that Probit models can be derived from an underlying

latent variable model where y * is an unobserved, or latent, variable, determined by:
(5.10) y: = x’iB + 8i, yi = Hy: > 0], 0 Otherwise
where the notion of 1[y: > 0] is used to define a binary outcome. The function

Yi = 1[y: > 0] is called the indicator function which takes on the value one if the event in
the brackets is true, and zero otherwise. Therefore, y is one if y* > O , and y is zero if
y* S 0. It is further assumed that e is independent of x, 8 has a standard normal

distribution, and 8 is symmetrically distributed about zero implying that

l— G(—z) = G(z) for all real numbers 2. Given the assumptions presented above, the

derived response probability for y is:

47

(5.11) Prob(y =1|x)= Prob(y* > O | x) = Prob[e > —(BO + xB) | x]
= 1- GKBO + x3)] = G(Bo + X13).

which is exactly the same as the binary response model shown in equation (5.7)

(Wooldridge, 2003). In this model the primary goal is to explain the effects of the x J- on

the response probability Prob(y = l Ix). In the application of these Probit models sample

selection bias and/or endogeneity bias may need to be dealt with, so depending on the

question at hand, different techniques may need to be applied.

5.2.3 Endogeneity Bias

An example of potential endogeneity bias in this analysis arises when estimating
whether producers believe that NAIS should be a mandatory system requiring all US.
cattle producers to participate. The interest lies in how various factors, such as
demographics and production/technology practices, affect producers’ beliefs regarding
mandatory traceability. Being currently registered in NAIS may be systematically
correlated with unobservable factors that affect beliefs about mandatory NAIS,
potentially producing biased estimators.

Endogeneity refers to the fact that an independent variable included in the model
(e.g., having NAIS registered premises) is potentially a choice variable, correlated with
unobservables relegated to the error term. In this situation, the dependent variable
(beliefs regarding if NAIS should be mandatory), is observed for all observations in the
data, so because the entire sample is used, there are no sample-selection issues (Millimet,
2001). Rather, this is an issue addressed in simultaneous-equation models considering

the potentially endogenous variables (having NAIS registered premises) in an equation

48

separate from the equation of original interest (beliefs regarding if NAIS should be
mandatory). Unlike single-equation models, in the simultaneous-equations models one
estimates the parameters of each equation while taking into account information provided
by other equations in the system (Gujarati, 2003). Following Gujarati (2003), consider
the following system of equations:

(51?) Yli = 1310 +YIIXIi tun

(5-13) Y2i = 1320 +521Y11+Y21X1i+u2i
where Y; and Y2 are mutually dependent, or endogenous, stochastic variables, X; is an
exogenous variable, and ul and u2 are stochastic disturbances terms. Thus, for
estimation purposes, it must be determined if the stochastic explanatory variable Y1 in
equation (5.13) is distributed independently of u] . Referring to equation 5.12 and 5.13

for a procedure on how to remedy this issue, first equation 5.12 is estimated and the
predicted values are obtained. Using the predicted values the Inverse Mills ratio is
calculated. Then, including the Inverse Mills Ratio as an explanatory variable in the
estimation of equation 5.13 (instead of Y1 itself) can be done to test for endogeneity. If
the Inverse Mills Ratio is found to be statistically significant, then evidence of
endogeneity exists, justifying the bias correction procedure of including the Inverse Mills
Ratio as an explanatory variable. If it is statistically insignificant, then we fail to reject

the null hypothesis of exogeneity.

49

5.2.4 Trinomial Probit

Extending this discussion in analyzing trinomial situations with the Probit model
we look to calculate the probability of choosing one of three alternatives. This can be
Shown by:

(5.14) Prob[U3 2 U1 and U3 2 U2] = PI’Ob[U1—U3 SOflI‘ld U2 —U3 SO]

5.2.5 Order Probit

An extension of the Probit model applies to models in which there is an ordering
to the categories associated with the dependent variable (Pindyck and Rubinfeld, 1996).
That is, often the response variable can have more multiple outcomes and very often the
outcomes are ordinal in nature; that is they cannot be expressed on an interval scale

(Gujarati, 2003). An example in this research is that some responses are on a Likert-type

’9 ‘6 3, 6‘

scale, such that a respondent indicates “entirely unimportant, unimportant, neutral,”

“important,” and “very important.” Within an ordered Probit model the following

specification was used:
* I
(5.15) yi = xii} + 3i
t . . . , , .
where VI 18 the latent and contInuous measure of Interest faced by respondent 1, Xi IS a

vector of explanatory variables describing respondent i, B is a vector of parameters to be

estimated, and Si is a random error term (assumed to follow a standard normal

distribution). The observed and coded discrete continuous measure of interest, y; , is

determined from the model as follows where it is assumed that there is an underlying

index Z for each respondent that measures the degree of their response. As shown in

50

Pindyck and Rubinfeld (1996) the ordered Probit model assumes that there are cut-off

points Z * and Z ** which define the relationship between the observed and unobserved
dependent variables, specifically, Zi = or + BXi, and
l3 if 2i 2 z”

(5.16) yi< 2 if z” < zi < z"
lif z, 52*
l

 

where the Zi’s represent thresholds to be estimated (along with the parameter vector [3).

This general specification may be extended to use in multiple size Likert-type response

questions.

5.3 Choice Model Specifications

Random utility theory frequently underlies objectives related to choice
experiments. Thus, models based on random utility, can be used to identify the set of
feasible alternatives producers may choose among a set of choices. As shown in

Nakosteen and Zimmer (1980) suppose that, an agent’s utility of two choices, can be

denoted Ua and Ub. The observed choice between the two reveals which one provides

the greater utility, but not the unobservable utilities. Thus, the observed indicator equals

1 if U3 > Ub and 0 if U21 S. Ub. There are multiple approaches that may be employed
to model random utility. To begin, a common formulation is the linear random utility

model (Greene, 2000):
(5.17) Ua = x'p, +2, and Ub = x'Bb +eb

Then, if Y = l is denoted by the agent’s choice of alternative a:

51

(5.18) Prob[Y = 1 | X] = Pr ob[Ua > Ub]
= Prob[x'Ba + ea -x’Bb - 8b > O | x]
= Prob[x'o. 431,) + 2. -eb > 01x1
= Prob[x'B + e > 0| x]

5.3.1 Multinomial Logit

This random utility framework can be applied to the initial sets of models that will
be estimated which are typically referred to as multinomial (or conditional) Logit models.
The conditional logistic models estimate producer random utility [Adamowicz et al.
(1998); Lusk, Roosen, and Fox (2003); and Schroeder et al. (2005)] which can be

characterized by the following equation:

where U j, is the utility associated with alternative j in choice scenario t, vJ-t is the

systematic, observable component of utility determined by attributes and their values, and

ejt , is a random, unobservable component of Logit models, independently and

identically distributed over all alternatives and choice situations. This random
component of the producer’s utility function is included to capture the variation in
producers choices (i.e., a producer may not choose what seems to the analyst to be the
“preferred” alternative or the analyst may simply fail to incorporate all relevant

explanatory variables in X). A producer will choose alternative j if U J- 2 Uk for all

j ¢ k. So, the probability that alternative j will be chosen is equal to the probability that

the utility gained from its choice is greater than or equal to the utilities of choosing

another alternative in the choice set. However Since these utilities contain a stochastic

52

component, researchers can only describe the probability of producers choosing

alternative j as (Boxall and Adamowicz, 2002; Adamowicz et a1, 1998):

(5.20) Prob(jchosen}= prob{vj + ej 2 vk +sk; j¢ k V j e C}

where C is the choice set of all possible alternatives. Assuming the random errors in
(5.19) are independently and identically distributed across the j alternatives and N
individuals with a type I extreme value distribution, Adamowicz et a1 (1998), Boxall and
Adamowicz (2002) and Lusk, Roosen, and Fox (2003) have shown that the probability of
a producer choosing alternative j becomes:

CHBXI

Eieufixk

keC

(5.21) Prob(i chosen}=

where u is a scale parameter, which is inversely related to the variance of the error term.
According to Lusk, Roosen, and Fox (2003) the scale parameter, p, is typically assumed
equal to one because it is unidentifiable within any particular data set. [3 is a vector of

parameters. Assuming the systematic utility component v j is linear in the parameters
and follows the generalized regression specification leads to:

(5.22) vj = 51le +132sz +... +anjn
where x jn is the n-th attribute value for alternative j and [in is a vector of preference

parameters associated with the n-th attribute of the j-t‘h alternative. The 13's are utility

parameters to be estimated, initially assumed to be constant across producers. That is
utility levels are independent of characteristics and perceptions that may vary across
producers. Multinomial Logit models assume that all respondents share the same

coefficients for a given attributes. That is, all respondents are assumed to have the same

53

preferences for attributes. This assumption may be unrealistic if producer’s tastes are in
fact heterogeneous. As such, the homogeneous producer tastes assumption is evaluated

using random parameter Logit and latent class models as described below.

5.3.2 Random Parameter Logit

Utility parameters can be allowed to vary across the sampled observations (as
random parameters) and therefore deviate from the surveyed population mean. A random
parameters Logit model, as well as the previously discussed models, was used to
determine producer willingness-to-accept (WTA) in alternative voluntary traceability
systems relative to one another. The RPL model allows for random taste variation within
the surveyed population, is free of the independence of irrelevant alternatives (IIA)
assumption, and allows correlation in unobserved factors over time, thus eliminating
three limitations of standard Logit models (Revelt and Train, 1998; Train, 2003; Tonsor
et al., 2005). These three limitations are also avoided in the latent class model shown
below. This aids in the ability to directly estimate heterogeneity in producer preferences.

Specification of the random parameters Logit model is given by:

(5-23) Uijt = Vijt + [uij + 81th
where Uijt is the utility producer i associates with attribute j in choice scenario t, Vijt is
the systematic portion of the utility firnction, uij is an error term distributed normally
over producers and alternatives (but not choice situations), and Sijt is the stochastic error,

independently and identically distributed over all producers, attributes, and choice
scenarios. This describes a panel data model where the cross-sectional element is

producer i and the time series component is the choice scenario j (Alfnes, 2004; Tonsor et

54

al., 2005). The probability that producer i chooses alternative j in choice scenario t is
given by:

(5.24) Prob(UiJ-t 2 Uikt)

for all possible k attributes. Following Alfnes (2004) and Tonsor et al. (2005) and

assuming Vijt is linear in parameters, the utility function can be expressed as:

(525) V1: = 311x“: + Bizxzit + + Pijxijt
where xijt is the j-th attribute value for choice scenario I for producer i and Bj is a vector

of preference parameters associated with the j-th attribute of the t-th choice to scenario of

the i-th producer. We specify 13 to vary normally across producers.

The random parameters provide a rich array of preference information. They
define the degree of preference heterogeneity through the standard deviation of the
parameters and through interactions between the mean parameter estimate and the
deterministic segmentation criteria, where the latter include other attributes of
alternatives, socio-economic and contextual descriptors, and even descriptors of the data
collection method and instrument. Random parameters are also the basis for
accommodating correlation across alternatives and across choice situations (Hensher,
Rose, and Greene, 2005).

RPL and latent class models examine heterogeneity differently and differ in their
ability to explain the identified heterogeneity. RPL models explicitly account for
heterogeneity by allowing parameters to vary randomly over individuals (Layton, 1996;
Train, 1997, 1998). While these random parameter Logit procedures incorporate and

account for heterogeneity, they are not well—suited to explaining the sources of

55

heterogeneity (Boxall and Adamowicz, 2002). These sources often relate to the
socioeconomic characteristics and tastes of the decision maker (Ouma et al., 2007).2
Latent class models are better suited in explaining the sources of heterogeneity, because
individuals are intrinsically sorted into a number of latent classes (Ouma et al., 2007).
Each segment (or latent class) is characterized by homogeneous preferences though

heterogeneous across classes (Boxall and Adamowicz, 2002).

5.3.3 Latent Classification Model

Boxall and Adamowicz (2002) describe the latent classification/segmentation
approach to assume the existence of s segments in a population where individual n
belongs to segments (5 = l, ..., s). The latent classification approach assumes that 3
segments of producer preferences exist such that preferences are homogeneous for
producers within each segment but heterogeneous across segments. The utility function
can now be expressed:

(5-26) Vnils = Bsxni +8nils

In this expression the utility parameters are now segment specific. The probability of a
producer choosing alternative j is:

ellsl3sxi

261155st
kEC

(5.27) Prob{/' chosenls}=

 

where the [3's and u's are segment-specific utility and scale parameters respectively.

 

2 Though it is possible to account for the socioeconomic characteristics of the decision maker by interacting
key individual characteristics with the traits, this requires a priori selection of key limited individual-
specific variables (Ouma et al., 2007).

56

The paper applies this latent classification approach to a set of voluntary beef
traceability system choice data. The behavioral components come from a choice
experiment in which three attributes of voluntary traceability systems were varied. The
analysis assesses simultaneously the influence of individual characteristics, motivational
aspects, and the influence of choice-based attributes in the estimation of latent segments

(Boxall and Adamowicz, 2002).

57

CHAPTER 6: EMPIRICAL MODELING

The survey data, which was summarized in chapter 4, was utilized to provide
economic insight into why varying types of cow-calf operations choose alternative
voluntary traceability systems and to identify those characteristics affecting producers’
choices. Appendix 2, as explained in Chapter 4, provides a complete list of the variables
elicited from these questions and complete definitions of each. This chapter follows a
structured dialogue of theoretical and empirical evaluation procedures used in examining
the appropriateness of the developed models. In particular, specific applications of Tobit,
Probit, and Logit models uniquely specified to address the objectives of this research
were presented.

The discussion of Tobit and Probit models will focus on the portion of the survey
not involved with the choice experiment. Models throughout this discussion will take the
following general form:

(6.1) Variable of interest = f(Demographics, Production Practices, Perceptions)
where Demographics, Production Practices, and Perceptions are vectors of multiple
variables. Discussion of MNL, RPL, and LCM models involved estimation of models
analyzing the choice experiment responses while also utilizing select variables from the
previously discussed portion of the survey in the analysis.

Producer’s demographics, production practices, and perceptions will likely have
large impacts on their decisions concerning their beliefs, current practices, and choices
regarding traceability systems. Controlling for demographics is necessary for examining
the relative impacts of demographics to the impacts of other factors like perceptions and

current production practices. Some production methods may decrease support because

58

producers would not like to share that information with the rest of the production chain;
while some production methods may increase support because producers can use these
claims to increase marketability. Some marketing methods may decrease support
because producers may not see the need when they are marketing directly to consumers
because the beef is already “traceable.” While other marketing methods may increase
support because producers can use these claims to increase marketability. Producers’
perceptions regarding important issues to the US. beef industry when designing a
voluntary traceability system examined if producers are more concerned with issues such
as disease implications or marketability. Perceptions regarding concerns to the US. beef
industry when designing a voluntary traceability system may also impact support.

A factor analysis was performed on three sets of producer perception variables or
eighteen statements (questions 17, 18, and 19 in the survey, Appendix 1) prior to
estimation. The scores from the 18 statements were factor analyzed using principle
component analysis with varimax rotation. Components were extracted until eigenvalues
were less than or equal to 1.0 (Boxall and Adamowicz, 2002). The factor analysis
identified three components across the given set of statements. However, these three
components were not very informative as they simply consisted of each question (17, 18,
and 19, respectively) within the survey. Therefore, the factors were not used as

explanatory variables in subsequent models.

6.1 Tobit Models

The survey asked cow-calf producers to forecast future NAIS registration and

RFID use rates among cow-calf producers in the US. beef industry. Examining the

59

characteristics that describe the most optimistic and pessimistic forecasters may provide a
better sense of how the NAIS traceability system and RFID technology could be
promoted to enhance voluntary participation. This analysis helps in meeting the
objective related to cow-calf producer forecasts in chapter 3.

More specifically, Tobit models will be used to evaluate responses to the two part
question (question 24 of the survey, Appendix 1):
0 ”If the NA IS system remains a system of voluntary participation, by December 31,
2008 what do you predict will be the percentage of US. cow-calf operations with
premises registered in NAIS? ”
0 “If the NAIS system remains a system of voluntary participation, by December 31,
2008 what do you predict will be the percentage of US. cow—calf operations with
premises registered in NAIS M using NAIS approved RF ID animal
identification devices on the percentage of cattle leaving their premises. ”
The forecasting questions will help give an indication of where producers believe the
future (in this case specified as December 31, 2008) of the beef industry is going with the
voluntary NAIS system. If a producer believes that NAIS registration and RF ID use rates
will be high in the future then arguably they may be more likely to accept and support
such programs. Alternatively, if they believe traceability will not be implemented within
the industry a lack of support may be more likely.

The bivariate empirical Tobit model for the forecast of NAIS participation and

forecast Of NAIS participation and RFID use took the form:

60

(6.2)

y: = a0 + otlAge + (12 Member + a3 Years + a4Eprears + a5Cows + a6Auctions
+ a7RFID + as Re gNAIS + orgManNAIS + (110C Retailers + 0111C Pr ocessors
+ or] ZCFeedlotS + e]

y; = so + BlAge + BZCB + B3NP + B4NW + B5SE +1363P + B7cMember + BgEducation
+ 139 Years + BloEprears + Bl llncome + Blelncome + [313Labor + B14OFeed

+ B15Cows + B16Auctions + [317PTags + Bl 3ENotches + [319Brand + Bonattoo
+ [32 lMTags + BZZRFID + [323NOID + [324 Re gNAIS + BZSManNAIS + 82

where y] = producer forecast of NAIS premises registration, y 2 = producer forecast of
NAIS premise registration and RF ID use, or' s and [3' s are coefficients to be estimated and
the explanatory variables are defined as described in Appendix 2. If the model’s error

terms are correlated, joint estimation will allow for an increase in effiency, thereby

leading to more consistent 13' s.

6.2 Probit Models

Probit models will be applied to these specific survey questions (22 and 23 in the
survey, Appendix 1):
(1) “Are your operation ’s premise(s) currently registered with USDA in the NAIS
Wational Animal Identification System)?” and
(2) “Do you believe that NAIS (as previously outlined) should be a mandatory

system requiring all US. cattle producers to participate?”

61

6.2.1 Binary Probit
Question (1) will follow a binary format as the question elicited a “yes or no”
response. The empirical Probit model for the first question of whether producers’

premises were currently registered in the NAIS system was:

(6.3)
yf = so + BlAge + BZCB + B3NP + B4NW + [358E + eésP + B7Member + BgEducation
+ [39 Years + BloEprears + [51 llncome + BleIncome + [313Labor + [314OFeed
+ B15Cows + B16Auctions + BI7PTags + B] 8ENotches + BlgBrand + Bonattoo
+ BZIMTags + B22RFID + B23NOID + B25ManNAIS + a]

where y1 = 1(0) if a producer answered Yes(No), [3' s are coefficients to be estimated,

and the explanatory variables are defined as described in Appendix 2.

6.2.2 Endogeneity Evaluation

Question (2), whether NAIS should be mandatory, followed a trinomial format as
the question elicited a “yes or no or undecided” response. To examine if responses to
question (2) were endogenously determined with question (1), a two-stage estimation was
performed (Wooldridge, 2002). Significance of the Inverse Mills ratio coefficient led to

reject exogeneity of question (2) from question (1).

6.2.3 Trinomial Probit

Estimating the trinomial Probit model was based upon producers’ beliefs
concerning mandatory NAIS; where producers chose between (i) NAIS §_I_1_O_ufl be
mandatory, (ii) undecided whether NAIS should be mandatory, and (iii) NAIS should not

be mandatory.

62

(6.4)
yilt = a0 + otIAgeit + or2CBit + 013NPit + a4NWit + aSSEit + a6SPit + 0L7Memberit
+ agEprears it + a9 Years it + alolncomeit + or] lCowsit + onleuctionsit
+ a13B Re tailersit + orMB Pr ocessorsit + or] 5BFeedlotsit + a16C Re tailers it

+ (117C Pr ocessorsit + or] 3CFeedlotsit + otlglnvMillsit + a},

Yizt = Bo + BIAgeit + [32081: + B3NPit + B4Nwit + 135351: + 13651’1: + B7Memberit
+ [33 Eprears it + [39 Years it + Blolncomeit + B1 lCowsit + [ileuctionsit
+ [313B Re tailersit + [314B Pr ocessorsit + B1 5BFeedlotsit + [316C Re tailersit

+ B17C Pr ocessorsit + B18CFe€dIOtSit + BlglnvMillsit + 8%

where yilt = 1 indicates a belief that NAIS should be mandatory (0 otherwise), and yizt = 1

indicates a producer being undecided about NAIS being mandatory (0 otherwise).3 The
Inverse Mills ratio (InvMills) is equal to the standardized predicted values from equation

6.3. [5' S are coefficients to be estimated, and the explanatory variables are defined in

Appendix 2. Maximum Likelihood Estimation (MLE) to jointly estimate equations (6.3)
and (6.4) would have been preferred; however, the PROC QLIM procedure in SAS
(which was used on the previous models and some of the subsequent models) does not
currently support multinomial Probit estimation. SAS does not provide an option to
specify multinomial Probit models for this system (SAS, 2008). Multinomial Probit
models are supported in the PROC MDC procedure in SAS; however, given the
endogeneity issue with the registered NAIS equation it does not allow systems of
equations to be estimated as would need to be done. Therefore, STATA was used in the
estimation of this equation and a two-stage procedure using the Inverse Mills ratio was

used to deal with the endogeneity issue.

 

3 y ilt = 0 and yizt = 0 implies a producer believes NAIS should not be mandatory.

63

6.2.4 Ordered Probit

Ordered Probits were used to estimate questions in which there was ranked
responses as given in the following questions (l7, l8, and 19 in the survey, Appendix 1):

(1) “In designing a national, individual animal traceability system how important
are the following issues in the US. beef industry (please circle your answers
where I = Entirely Unimportant, 2 = Unimportant, 3 = Neutral, 4 =
Important, 5 = Very Important)?”

(2) “In designing a national, individual animal traceability system how concerned
are you regarding the following issues in the US. beef industry (where I =
Entirely Unconcerned, 2 = Unconcerned, 3 = Neutral, 4 = Concerned, 5 =
Very Concerned)?”

(3) “Indicate your level of agreement with each of the following statements

(Where 1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 =
Strongly Agree). Implementing individual animal traceability systems: (i) “is
more cost eflective for larger cow-calf operations. " (ii) “results in more
liability for cow—calf producers than cattle owners at other stages of
production ” (iii) “is unnecessary if COOL (C ountry-of-Origin Labeling) was
implemented nationally. ” (iv) “as a mandated system is exaggerated?”

The empirical Probit model for these questions was:

(6.5)
y: = BO + BlAge + BZCB + B3NP + B4NW + [358E + B6SP + B7Member + figEducation
+ BgYears + BIOEprears + 131 llncome + Blelncome + [313Labor + BMOFeed

+ B] 5Cows + [316Auctions + B17 Re gNAIS + 81

64

where [3' S are coefficients to be estimated, and the explanatory variables are defined as

described in Appendix 2.

6.3 Multinomial Logit Models

The choice experiment framework was designed to improve the understanding of
traceability system choices and preferences by requiring cow-calf producers to choose
between two traceability system options and a No Traceability system option (reference
data/survey discussion and/or Appendix A showing the CE scenarios). Data obtained
through choice experiments has traditionally been analyzed using multinomial Logit
models (e.g., Lusk, Roosen, and Fox, 2003).

Because, producer-level support, or lack of support, for voluntary traceability
systems in the US. will be based on the net benefits producer groups perceive they would
receive from the system and their underlying utility functions, empirical modeling and
subsequent examining of preferences for alternative systems should allow researchers to
better determine optimal voluntary systems (Bailey and Slade, 2004).

Recall that multinomial logistic models estimate producer random utility as

(Adamowicz et al., 1998; Lusk, Roosen, and Fox, 2003; and Schroeder et al., 2005):

(6.6) th = Vjt +8jt
where U jt is the utility associated with alternative j in choice scenario t, v jt is a

systematic, observable (explainable) component of utility determined by attributes and

their values, and sjt , is a random, unobservable (unexplainable) component of Logit

models, independently and identically distributed over all alternatives and choice

65

situations. Assuming vJ-t is linear in parameters, vJ-t is specified individually for each of

the three available alternatives (two traceability system options and a No Traceability
system option):

(6.7) vj, = B1(PDJ-t) + B2(Pvtljt) + B3(PvtNIJ-t) v j = NAIS Trace

(6.8

Vjt )= BAdv Trace + 131(PDjt) +132(PVfljt)+ 133(PV1N1jt) + 134(PPjt ) + 55(PGjt)
+156(HRjt)+ 137(AVPPjt) + B8(AVPGjt) + B9(AVHRjt) + BlO<PPPGjt)
+ BI 1(PPHRJ-t) + B12(PGHRjt) + BI3(AVPPPGJ-t) + BI4(AVPPHRjt)
+ B15(PPPGHRJ-t) + B16(AVPPPGHRJ-t) + B17(AT _Cows) + B18(AT _ Member)
+ B19(AT _ Auctions) + B20(AT _ Yearsjt) + B2] (AT _ IndID)
+ B22(AT __ GrpIDJ-t) + B23(AT _ NOID) + B24(AT _ CB) + B25(AT _ NP)
+ B26(AT _ NW) + B27(AT_SE) + B23(AT _SP) V j = Adv Trace

(6.9)
vjt = BNO Trace + 81(PDjt) + 629(NT _ COWS) + [330(NT __ Member)

+ B31 (NT _ Auctions) + B32(NT _ Years) + B33(NT _ IndID)

+ B34(NT _ GrpID) + B35(NT _ NOID) + B36(NT _ CB)

+ B37<NT _ NP) + 1333(NT _ NW) + 1339(NT_SE)+ 1340(NT_SP)
V j = No Trace

where B Adv Trace and BN0 Trace are constants (relative to NAIS) for Advanced
Traceability and No Traceability, respectively. PD jt is the premium/discount per head
sold; Pvtljt and PvtNIJ-t are effects coded variables equal to one if the alternative lists

the entity maintaining the data as Private - Industry or Private - Not Industry, respectively

(-1 for Government (base) or 0 otherwise); PPJ-t, PG jt , HR jt, AVPPJ-t , AVPGJ-t,

AVHRjt, PPPGjt , PPHR AVPPPGjt , AVPPHRJ-t , AVPGHRjt , PPPGHRjt,

Jt’

AVPPPGHR jt denote effects coded variables equal to one if the alternative lists the

66

additional information as Production Practices, Performance/Genetics, and Health
Records and for combinations refer to Appendix B, respectively (-1 for Age Verification
(base) or 0 otherwise). Interaction terms AT_Cows, NT_Cows, AT_Member,
NT_Member, AT_Auctions, NT_Auctions, AT_Years, NT_Years, AT_IndID,
NT_IndID, AT_GrpID, NT_GrpID, AT_NOID, NT_NOID, AT_CB, NT_CB, AT_NP,
NT_NP, AT_NW, NT_NW, AT_SE, NT_SE, AT_SP, and NT_SP are explanatory
variables, that incorporate producer demographic effects on preferences for Advanced
and No Traceability systems, and are defined in Appendix 2.

As explained above the managing entity and additional information variables
were effects coded. Following Beck and Gyrd-Hansen (2005), attributes with L
qualitative levels are transformed into L-l dummy variables in which each dummy is set
equal to 1 when the qualitative level is present, and equal to -1 if the Lth (the arbitrary
reference level) is present and equal to 0 otherwise. Thus, for our effects coding, the base
case for managing entity will be government and for additional information the base case
will be age verification. In effects coding, the reference point is defined as the negative
sum of the estimated coefficients; the utility of the Lth level equals:
Bl x (—l) + B2 x(—1)+---+ B1,, x (—l). This means that the reference point is now
internalized in the B estimates and cannot be carried over onto the B Adv Trace and

BN0 Trace coefficients. Thus, when effects coding is applied the constant term can only

reflect the utility associated with the fixed comparator and misinterpretation is not
possible (Beck and Gyrd-Hansen, 2005).
Standard multinomial Logit models may be limiting because they do not allow for

random taste variation, unrestricted substitution patterns, and correlation in unobserved

67

factors over time. Therefore, additional models were estimated including RPL and LCM

models.

6.4 Random Parameter Models

Random parameters Logit models explicitly account for heterogeneity by
allowing model parameters to vary randomly, and continuously, over individuals (Layton
1996; Train 1997, 1998). Although heterogeneity is accounted for in these models,
difficulty arises in determining the sources of heterogeneity (Ouma, Abdulai, and

Drucker, 2007). Assuming the observable portion of utility is linear in parameters, we

specify Vijt separately for each of the three available alternatives (two traceability system

options and a No Traceability system Option):

(6.10) Vijt = B] (PDjt) + [52(PVIIjt) + B3(PVINIjt) V j = NAIS Trace

6.11
(Vijt =)l3iAdv Trace + B1(PDjt) + 132(PV11jr)+l33(PVtN1jt)+ 134(PPjt) + 55(PGjr)
+136(HRjr)+ 137(AVPPjr) + B8(AVPGjt) + 139(AVHRjt) + 1310(PPPGjt)
+ B11(PPHRjt) + B12(PGHRjt) + B13(AVPPPGjt) + B14(AVPPHRJ-t)
+ B15(PPPGHRJ-t) + B16(AVPPPGHRJ-t) + B17(AT_Cowsi ) + B13(AT _ Memberi)
+ B19(AT _ Auctionsi) + B20(AT _ Yearsi ) + B21(AT _ IndIDi) + B22(AT __ GrpIDi)
+ B23(AT_NOIDi) V j = Adv Trace

Vijt = BiNo Trace + BI (PDjr) + 1324(NT _ COWSi) + 1325(1)]T _ Membefi)
+ B26(NT _ Auctionsi) + B27(AT _ Yearsi ) + B28(NT _ Yearsi)
(6.12) + B29 (AT __ IndIDi) + B30(NT _ IndIDi) + B31(AT _ GrpIDi)
+ B32 (NT _ Grp.IDi) + B33(AT _NoIDi) + B34 (NT_ NOIDi)
V j = No Trace

where the variables in equations (6.10), (6.11), and (6.12) are defined as in equations

(6.7), (6.8), (6.9). In addition, Pi Adv Trace and BiNo Trace are now assumed to be

68

normally distributed (B ~ N(B,Q)), thus introducing preference heterogeneity across

producers into the model. Revelt and Train (1998) define Pik on these two constants as:
(6.13) Br. = Br + LM
where L is a lower-triangular Cholesky factor of Q such that LL' = Q , and M is a vector
of independent standard normal deviates (Revelt and Train, 1998). Schroeder et al.
(2006) have shown that the RPL model estimates both the mean and standard deviation of
the utility coefficients assumed to vary across producers. Conversely, the MN L model
only estimates a mean coefficient across individuals and implicitly assumes this
coefficient adequately reflects each individual’s preferences. If the standard deviation
parameters are significant, evidence is said to exist supporting preference heterogeneity

for the evaluated attribute(s).

6.5 Latent Classification Models

A latent classification strategy assumes that producers can be separated into
different segments such that producers within each segment have homogeneous
preferences, but that preferences vary across producers from alternative segments. This
process may help better identify marketing segments of producers likely to support
alternative traceability systems. Observed variables (e. g., producer attitudes, perceptions,
socio-demographic, and/or production practice factors) are integrated with information
from the choice experiment such that some sources of heterogeneity can be identified
while simultaneously modeling revealed preferences for the traceability system attributes
in the choice experiment. The LCM model estimates a utility function unique to each

producer segment. Thus, this process groups producers with relatively homogeneous

69

preferences based on similarities in their utility functions, as revealed by a discrete set of

different values of each Bk from the MN L models. As shown below, additional

information variables were only included as singles and pairs due to convergence
problems with larger combinations of these variables. Assuming the observable portion
of utility is linear in parameters and that the traceability system attributes are additive

above the pair combinations, we Specify Vijt separately for each of the three available

alternatives (two traceability system options and a No Traceability system option):

(6.14) Vijtls = Bil|s(PDjt) + Bi2|S(PvtIJ-t) + Bi3|S(PvtNIJ-t) V j = NAIS Trace

(6.15)

Vijt|s = PAdv Trace + l3i1|s(PDjt)+Bi2|s(PVthr)+ 1313|s(PVtN1jr) + Bi4|s(PPjt) + BiSls(PGjt)
+ Bi6|s(HRjr) + Bi7|s(AVPPjt) + Bisls(AVPGjt) + Bi9|s(AVHRjt) + Bi10|s(PPPGjr)
+ Bil l|s(PPHRjt) + BIIZIS(PGHRjt) V j = Adv Trace

(6-16) Vijtls = BNO Trace + Bills<PDjt) V J = N0 Trace

where the variables in equations (6.14), (6.15), and (6.16) are defined as in equations
(6.7), (6.8), (6.9) above. Furthermore, membership in segments is a function of Cows,

Auctions, Member, Years, IndID, GrpID, and NoID. These variables are defined in

Appendix 2.

6.6 Willingness-to-Accept Estimates

MNL, RPL, and LCM model estimated coefficients themselves have little
interpretive value. However, in each model relative combinations of select coefficients
used for WTA estimates provide economically meaningful insights on producer
preferences. Following Nahuelhaul, Lourero, and Loomis (2004) and Rigby and Burton

(2005 ) mean WTA for producers are calculated for each non-price traceability system

70

attribute at the means of each model’s explanatory variables. These calculations are
generally given by.

MU

(6-17) Mean WTA = m lmean independent variables

where MU is the mean marginal utility of traceability systems and system attributes and
M U] is the marginal utility of income (proxied in our calculations with the
premium/discount (PD) price coefficient).

In estimating our RPL model, not allowing the price coefficient to to vary
randomly ensures a negative proice coefficient for all producers and ensures that WTA
estimates are normally distributed (Lusk, Roosen, and Fox, 2003).

To determine if the estimated average WTA estimates in each model are
statistically different from zero, a Krinsky-Robb (1986) bootstrapping procedure was
employed to develop confidence intervals on the WTA estimates. Bockstael and Strand
(1987) have emphasized that the parameter estimates used to calculate welfare measures
are themselves random variables, thus the Krinsky-Robb technique used here to develop
confidence intervals for the WTA measure accounts for the variability associated with all
the estimated coefficients and is based directly on the Logit specification for the choices

of respondents. The Krinsky and Robb approach uses the information on the distribution

of B contained in the variance-covariance matrix to approximate the distribution of WTA

(Park et al., 1991).
Following Park, Loomis, and Creel (1991), the Krinsky-Robb technique can be

implemented using information readily available from the estimated Logit model: the

estimates of the parameter vector, denoted by B, and the estimated variance-covariance

71

matrix, denoted by V . Multiple random drawings to create a new parameter vector Bare

made from a multivariate normal distribution with variance-covariance matrix V and

mean B . For each drawing of B, WTA is calculated. An empirical distribution for each
WTA measure is then obtained using the complete set of replications (random drawings).
A (1 — or) confidence interval is obtained by ranking the vector of calculated WTA values
and dropping the or / 2 values from each tail of the ranked vector (Park, Loomis, and
Creel, 1991). Krinsky and Robb (1986) suggest that past experimentation has indicated
that one-thousand drawings are sufficient to generate a sufficiently accurate empirical
distribution. This simulation process makes more complete use of valuable information
provided by each estimated model and results in much more complete mapping of

producer preferences (Tonsor et al., 2008).

6.7 Welfare Measure

In meeting one of the previously stated tier 1 objectives the welfare effect
producers would experience given two alternative changes were estimated: (1) removal
of No Traceability and (2) removal of both No Traceability and Advanced Traceability
from their choice set. The welfare measure accounted for the fact that producers are
currently free to choose “No Traceability” and that producers’ actual choices are
uncertain. Small and Rosen (1978) and Morey (1999) show that expected maximum
utility from making a choice from a particular choice set is given by: CV = lntevj)+ C,

where C is Euler’s constant. Thus, the welfare change that occurs when moving from

one situation given by CV0 to a situation given by CV1 is: l/(Marginal utility of

72

income) x [(CV1 )-(CVO )]. This calculation represents the most producers would be
willing to pay per choice occasion to face the choices in Situation 1 versus situation 0
(Lusk, Norwood, and Pruitt, 2006). Alternatively, one can interpret this value as the
amount necessary to make a producer equally well-off in the two situations. Chapter 7
will apply the modeling techniques described throughout this chapter to the data obtained

from the survey.

73

CHAPER 7: RESULTS

7.1 Producers’ Practices and Perceptions Regarding Traceability

The following chapter provides the econometric results and economic
interpretation of the various models estimated meeting the research objectives as
discussed in previous chapters. The end of this chapter includes parameter estimates and
standard error values for the various models estimated throughout this section. F -tests
were utilized to determine the preferred variable set for each model. F-tests allow for the
testing of joint hypothesis that a number of coefficients were zero. In order to determine
each preferred model, a complete set of producer specific variables were identified and
used in estimation of each model. Then groups of coefficient estimates were tested to
determine if the groups were statistically different than zero.4 Significance of individual
coefficients and mean marginal effects (ME) in each final model specification are
discussed throughout the text in this section. ME represent the mean effect of a particular
explanatory variable on the question/model at hand.

Additional models concerning producers’ perceptions and concerns when
implementing an individual animal traceability system were estimated, but not included
in this chapter. These include models of how important producers feel an individual
animal traceability system is for managing the supply chain, enhancing food safety,
improving marketability, and improving on-farm management. Other additional models
evaluated how producers perceived common issues in the beef industry when

implementing individual animal traceability systems (i.e., economies of scale, liability

 

4 The 10% significance level was used on all F-tests.

74

shifts, and the exaggeration in need of mandatory traceability). These results can be

found in Appendix 4.

7.1.1 Producers’ Forecast Analysis

Recall the two part question regarding producers’ forecast of NAIS participation
and both NAIS participation and RF ID use (question 24 of the survey, Appendix I).

(1) “If the NAIS system remains a system of voluntary participation, by December

31 , 2008 what do you predict will be the percentage of US. cow-calf
operations with premises registered in NA] .”

(2) “If the NAIS system remains a system of voluntary participation, by December

31, 2008 what do you predict will be the percentage of US. cow-calf
operations with premises registered in NAIS?”
The empirical equations to these questions were estimated jointly due to the possible
correlation between them due to potentially omitted covariates in each model. Estimating
a bivariate model identified significant correlation between the two equations. Therefore,
joint estimation was preferred.

The following are the demographics that were included in the Tobit equation to
evaluate producers’ forecast for NAIS participation: age (Age), the number of
memberships in various farm organizations (Member), years raising beef cattle (Years),
expected future years raising beef cattle (Eprears), number of beef cows that calved in
2007 (Cows), whether local auctions were used as a marketing outlet (Auctions), whether
RFID was used for animal identification (RFID), whether the operation’s premise(s) was

currently registered in NAIS (RegNAIS), producer’s belief on whether NAIS should be

75

mandatory (ManNAIS), and the percentage of costs that retailers, processors, and feedlots
born over cow-calf producers (CRetailers, CProcessors, and CF eedlots, respectively).
Variables included in the Tobit equation to evaluate producers’ forecast for NAIS
participation and RFID use were: Age, U.S. production regions (CB, NP, NW, SE, SP),
Member, education (Education), Years, Eprears, pre-tax household income (Income),
portion of household income from off-farm sources (Olncome), paid labor percentage
(Labor), the proportion of feed produced on-farm (OFeed), Cows, Auctions, whether
plastic ear tags (PTags), ear notches (ENotches), brands (Brand), tattoos (Tattoo),
brucellosis/metal tags (MT ags), RF ID, or no identification (NoID) were used for animal
identification, RegNAIS, and ManNAIS.

The mean response by respondents giving a forecast for future NAIS participation
was 40.47% with a standard deviation of 18.87%, while the mean response by
respondents giving a forecast for future NAIS participation and RFID use was 31.58%
with a standard deviation of 20.1 1%. Table 7.] presents the results of the significant ME
joint estimates for future NAIS participation and both future NAIS participation and
RF ID use.5 It is important to note that the ME are interpreted at the mean, thus for the
representative (average) producer in each case. This follows for all subsequent

interpretation of the ME coefficients.

 

5 Table 7.1a shows the estimated coefficients and standard errors. The remainder of this chapter follows
this format, where only the marginal effects of the covariates with statistically significant coefficients are
explicitly discussed.

76

Table 7.1. Producers’ Forecasts

 

 

Parameter ME

Future NAIS Participation

Years -1.6511

RegNAIS 9.7144

CProcessors -0.0821

CFeedlotS 0.0864
Future NAIS Participation and RF ID Use

Age 0.3487

OF eed -2. 1543

Cows -1.1635

RF ID -5.8898

ManNAIS 3.1342

 

0 Source: Calculated from Survey Data Estimation

Producers that have premises currently registered in NAIS and believe that more
of the costs from traceability systems are born by feedlot operations than by cow-calf
operations forecast higher NAIS premises registrations. Alternatively, producers with
more years of experience who believe that more of the cost from traceability systems are
born by processors than by cow-calf producers forecast lower NAIS premises
registrations.

Economic interpretation of the ME estimates reveals that producers with premises
registered in NAIS forecast a 9.71% higher NAIS premises registrations. For every
additional percentage of costs a producer perceives being born by processors they
forecast 0.08% lower NAIS premises registrations. For every five years of additional
experience producers forecast 1.65% lower premises registrations. For every additional
percentage of costs a producer perceives being born by feedlots over those born by cow-
calf producers, they forecast 0.09% higher NAIS premises registrations.

Older producers who believe NAIS Should be a mandatory system forecast higher
NAIS premises registrations and RFID use. While producers on larger operations whose

feed needs are primarily from on-farrn sources and who use RFID for animal ID forecast

77

lower NAIS participation and RF ID use. For every additional ten years of age producers
forecast 3.94% higher NAIS premises registrations and RF ID use. Operators who believe
NAIS should be mandatory will forecast 3.13% higher NAIS premises registrations and
use of RFID. Alternatively, for every level of beef cows that calved in 2007 increase (see
question 14 in survey, Appendix 1) producers will forecast 1.16% lower NAIS
participation and RFID use. Increasing the proportion of feed/forage needs that an
operation produces on their own farm by 25% is associated with a 2.15% lower forecast
for NAIS premises registrations and RF ID use. Producers who use RFID for animal ID
will forecast 5.89% lower NAIS participation and RF ID use.

Although obtaining and identifying these two sets of producer forecasts was
identified as one of the objectives; we do not place much value in these forecasts as
presented above. There are two primary reasons that lead us to this conclusion. First,
this question was the last question of the survey making answers more questionable.
Moreover, 9.31% of the producers failed to answer either part of this question.
Furthermore, of the completed questions, 9.47% of the answers were nonsensical as
producer forecasts for NAIS participation and RFID use were higher than only NAIS
participation. Thus, survey design and/or producer understanding (or survey completion

fatigue) led to the problems estimating, evaluating, and meeting this objective.

7.1.2 Registration in NAIS
The specific question addressing whether or not producers have their premises
registered in NAIS was: “Are your operation ’s premise(s) currently registered with the

USDA in the NAIS System? ” Variables included in the Probit model to analyze

78

producers’ premise registration in NAIS were: Age, CB, NP, NW, SE, SP, Member,
Education, Years, Eprears, Income, OIncome, Labor, OFeed, Cows, Auctions, PTags,
ENotches, Brand, Tattoo, MT ags, RFID, NoID, RegNAIS, and ManNAIS. Of the
respondents who responded to this question, forty-four percent of producers indicated
that their premises was currently registered in NAIS. Table 7.2 presents the ME

estimates for NAIS premise registration.6

Table 7.2. NAIS Premise Registration

 

 

Parameter ME

NW -0.3084
Member 0.0570
RF ID 0.4488
ManNAIS 0.0667

 

0 Source: Calculated from Survey Data Estimation

Here, producers who are more active in farm organizations, use RFID on their
premises, and believe NAIS should be mandatory are more likely to have their
operation’s premises registered in the NAIS system. However, producers from the
Northwest (relative to the Northern Crescent) are less likely to have their premises
registered with NAIS.

Each additional membership in a farm organization increases the likelihood of
producers registering their premises by 5.70%. RFID use on a premise increases the
likelihood of NAIS premise registration by 44.88%. Producers who believe NAIS should
be mandatory are 6.67% more likely to have premises registered. Finally, cow-calf
operators in the Northwest are 30.84% less likely than those in the Northern Crescent to

have the premises registered with NAIS.

 

6 Table 7.2a shows the estimated coefficients and standard errors.

79

7.1.3 Mandatory NAIS Beliefs

Next, we looked at the question: “Do you believe that NAIS should be a
mandatory system requiring all US. cattle producers to participate (please indicate yes,
no, or undecided). ” Variables in this discrete multinomial Probit model were: Age, CB,
NP, NW, SE, SP, Member, Education, Years, Eprears, Income, Cows, Auctions, the
percentage of benefits that retailers, processors, and feedlots receive over cow-calf
producers (BRetailers, BProcessors, and BF eedlots, respectively), CRetailers,
CProcessors, CF eedlots, and the inverse mills ratio (InvMills). Recall, the inverse mills
ratio (InvMills) is a standardized value representing the prediction/probability of a
producer having their premise registered with NAIS. Including this variable in the model
allowed us to deal with the inherent endogeneity between whether or not a producer had
the premises registered with NAIS and if they believe NAIS should be mandatory.

For the sample used in our model estimation, 21.22% of respondents believe
NAIS should be mandatory, 50.32% believe NAIS should not be mandatory, and 28.46%
of respondents are undecided concerning mandatory NAIS. Tables 7.3 presents the
results of the ME estimates for beliefs concerning whether NAIS should be mandatory
under the three possible outcomes (Yes, No, and Undecided).7 Here, ME’s of a given

covariate will by definition sum to zero across the given number of responses.8

 

7 Table 7.3a presents the estimated coefficients and standard errors.
8 While the MES sum to zero (and hence can be inferred fi'om Table 7.3) we have denoted MES associated
with statistically insignificant coefficients by N/A.

80

Table 7.3. Producers’ Beliefs Concerning Mandatory NAIS

 

 

 

ME
Parameter No Yes Undecided
Age 0.0015 N/A -0.0072
CB -0.3370 0.1549 0.1821
NP -0.5465 0.2565 0.2900
NW -0.6665 0.1801 0.4864
SE -0.4554 0.2410 0.2144
SP -0.4268 0.3267 0.1001
Member 0.1478 -0.0194 —0.1284
Years -0.0131 N/A 0.0474
Income 0.0391 N/A -0.0305
Auctions -0. 1565 N/A 0.2131
BRetailerS 0.0020 -0.0015 N/A
CFeedlots -0.0050 0.001 1 0.0039
InvMills 1.3550 -0.5089 -0.8461

 

0 Source: Calculated from Survey Data Estimation

Older producers that are more active in industry organizations, with higher levels
of household income, and beliefs that retailers, as opposed to cow-calf producers, receive
more of the benefits from traceability systems are more likely to believe NAIS should not
be mandatory. While, producers from the Corn Belt, Northern Plains, Northwest,
Southeast, and Southern Plains who are less active in industry organizations, and believe
feedlots incur more of the costs from traceability systems than cow-calf operations are
more likely to answer Yes or Undecided when asked if NAIS should be mandatory.
Furthermore, producers believing retailers receive less of the benefits from traceability
systems than cow-calf producers are more likely in favor of mandatory NAIS. Where as,
younger, more experienced producers with a lower level of household income who use
local auctions to market their cattle are more likely undecided concerning mandatory
NAIS

For every ten years a producer’s age increases they are 1.5% more likely to

believe NAIS Should not be mandatory and 7.2% less likely to be undecided. Producers

81

from the Corn Belt, Northern Plains, Northwest, Southeast, and Southern Plains, as
opposed to producers in the Northern Crescent are 33.70%, 54.65%, 66.65%, 45.54%,
and 42.68%, respectively, less likely to believe NAIS should not be mandatory; and
15.49%, 25.65%, 18.01%, 24.10%, and 32.67%, respectively, more likely to believe
NAIS should be mandatory. While these same producers are 18.21%, 29.00%, 48.64%,
21.44%, and 10.01%, respectively, more likely to be undecided concerning mandatory
NAIS. Each addition membership in a farm organization causes producers to be 14.78%
more likely to believe NAIS should not be mandatory, 1.94% less likely to believe NAIS
should be mandatory, and 12.84% less likely to be undecided. For every additional five
years of experience, producers are 1.31% less likely to be in favor of voluntary NAIS and
4.74% more likely to be undecided when asked if NAIS should be mandatory or not.

For every $25,000 increase in household income producers are 3.91% more likely to
believe NAIS should not be mandatory and 3.05% less likely to be undecided9.

Producers who believe more of the benefits from traceability systems translate to retailers
as opposed to cow-calf producers are 0.21% more likely to be in favor of no mandatory
NAIS.

Producers using local auctions to market their cattle are 15.65% less likely to be
in favor of voluntary NAIS and 21.31% more likely to be undecided concerning
mandatory NAIS. For every additional percentage of benefits a producer believes is
received by retailers over those received by cow-calf producers, they are 0.20% more
likely to be in favor of voluntary NAIS and 0.15% less likely to be in favor of mandatory

NAIS. Similarly, for every additional percentage of costs a producer perceives being

 

9 These results and interpretations are not very informative, but do suggest that a continuous variable may
have very small effects. In the future, it would be preferable to use a continuous variable to make
interpretation more useful.

82

born by feedlots over those born by cow-calf producers, they are 0.50% less likely to be
in favor of voluntary NAIS, 0.11% more likely to believe NAIS should be mandatory,

and 0.39% more likely to be undecided.

7.1.4 Important Issues when Implementing Traceability Systems

In meeting our second tier of objectives, we evaluated important issues within the
beef industry as they relate to traceability system implementation. As shown by
summary statistics presented in Appendix 3, cow-calf producers vary widely in scale and
production practices of their operations, so it is crucial to consider how producer specific
demographics affect perceptions and concerns of cow-calf producers. Controlling for
demographics and production practices is necessary for examining the relative impacts of
different factors on perceptions and levels of concerns. The following are the
demographics that were included in each model: Age, CB, NP, NW, SE, SP, Member,
Education, Years, Eprears, Income, OIncome, Labor, OFeed, Cows, Auctions, and
RegNAIS.

A series of discrete ordered Probits were estimated to evaluate how cow-calf
producer demographics and production practices influenced cow-calf producer’s
perceptions and concerns regarding a national, individual animal traceability system. It
was important to recognized the possible correlation between various questions due to
potentially omitted covariates in each model. As such, it would have been preferable if
these equations would have been estimated in a system with all equations being included,

however, convergence constraints and software limitations (note there were 18 ordered

83

Probit models) restricted our ability to do 30.10 Estimating bivariate models on pairs of
equations, subjectively identified to be potentially related, regularily identified significant
correlation between the two models. However, single-equation ordered Probit models
were settled upon as conclusions were not substantially changed as compared to
subjective bivariate specifications and convergence was feasible.

We examined responses to the following question, “In designing a national,
individual animal traceability system how important are the following issues in the US.
beef industry (please circle your answers where 1 = Entirely Unimportant, 2 =
Unimportant, 3 = Neutral, 4 = Important, 5 = Very Important).” This question was
asked with specific reference to the following issues: monitoring/managing disease,
maintaining current foreign markets, accessing foreign markets, and increasing consumer
confidence.” Appendix 3 (Tables 23-23h) presents a summary of producer’s answers to
these questions.

This summary reveals that cow-calf producers believe that the most important
issues to the US. beef industry in designing a national, individual animal traceability
system were monitoring/managing disease and increasing consumer confidence, with
over 78% of producers ranking these as important or very important. Maintaining current
foreign markets and accessing foreign markets were seen as important (important or very
important) as indicated by over 50% of producers. This shows that producers are dually
concerned with disease implications and marketability of their beef cattle when

considering the design of a traceability system.

 

'0 These eighteen ordered Probits involving the design of a national, individual animal traceability system
in the US. beef industry included: Eight for important issues, six for producer concerns, and four for
common industry statements.

84

Table 7.4 presents the ME estimates for how important producers feel traceability
is for monitoring/managing disease. “ The ME estimates were interpreted for the very
important response. Once again given the nature of these particular models, ME’S of a

given covariate will by definition sum to zero across the given number of responses.

Table 7.4. ME to Monitoring/Managing Disease

 

 

 

ME
Entirely Very
Parameter Unimpgrtant Unimportant Neutral Important Important
Age 0.0012 0.0006 0.0013 0.0012 -0.0042
Member -0.0086 -0.0044 -0.0090 -0.0084 0.0303
Education -0.0067 -0.0034 -0.0071 -0.0066 0.0239
RegNAIS -0.0329 -0.0168 -0.0345 -0.0322 0.1164

 

0 Source: Calculated from Survey Data Estimation

Younger producers of higher education, with more active membership in farm
organizations, and registered NAIS premises see greater importance (are more likely to
respond very important) for monitoring/managing disease in designing a national,
individual animal traceability system.

For every ten years a producer’s age increases they are 4.20% less likely to
respond that traceability is very important for monitoring/managing disease. Each
additional membership in a farm organization causes producers to be 3.03% more likely
to respond that traceability is very important for monitoring/managing disease. For every
level of education increase (see question 5 in survey, Appendix 1) producers are 2.39%
more likely to respond that traceability is very important for monitoring/managing
disease. Producers with their premises registered in NAIS are 11.64% more likely to
respond that traceability is very important for monitoring/managing disease.

Table 7.5 presents the ME estimates for how important producers feel traceability

 

” Table 7.4a shows the estimated coefficients and standard errors.

85

is for maintaining current foreign markets.12

Table 7.5. ME to Maintaining Current Foreign Markets

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
Age 0.0012 0.0006 0.0015 0.0004 -0.0038
Auctions 0.0337 0.0170 0.0415 0.0115 -0.1038
RegNAIS —0.0231 -0.01 17 -0.0284 -0.0079 0.0712

 

0 Source: Calculated from Survey Data Estimation

Younger producers not using local auctions as their most frequent marketing
outlet and with premises registered in NAIS see greater importance (are more likely to
respond very important) for maintaining current foreign markets. For every ten years a
producer’s age increases they are 3.80% less likely to respond that traceability is very
important for maintaining current foreign markets. Producers using local auctions as a
marketing outlet are 10.38% less likely to respond that traceability is very important for
maintaining current foreign markets. Producers with their operation’s premises
registered in NAIS are 7.12% more likely to respond that traceability is very important
for maintaining current foreign markets.

Table 7.6 presents the ME estimates for how important producers feel traceability

is for accessing foreign markets.'3

Table 7.6. ME to Accessing Foreign Markets

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
Education -0.0100 -0.0052 -0.0125 -0.001 1 0.0288
Years -0.0098 -0.0051 -0.0122 -0.0010 0.0281
Auctions 0.0369 0.0192 0.0459 0.0039 -0.1059

 

0 Source: Calculated from Survey Data Estimation

 

'2 Table 7.5a shows the estimated coefficients and standard errors.
'3 Table 7.6a shows the estimated coefficients and standard errors.

86

Producers with higher education, more years of experience, not using local
auctions as a marketing outlet see greater importance (are more likely to respond very
important) for accessing foreign markets.

For every level of education increase (question 5 in survey, Appendix 1)
producers are 2.88% more likely to respond that traceability is very important for
accessing foreign markets. For every five years of additional experience, producers are
2.81% more likely to respond that traceability is very important for accessing foreign
markets. Producers marketing through local auctions are 10.59% less likely to respond
that traceability is very important for accessing foreign markets.

Table 7.7 presents the ME estimates for how important producers feel traceability

. . . 4
18 for Increasrng consumer confidence.1

Table 7.7. ME to Increasing Consumer Confidence

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
Age 0.0013 0.0009 0.0014 0.0010 00046
SE -0.0695 —0.0456 -0.0718 -0.0552 0.2421
Auctions 0.0320 0.0210 0.0331 0.0255 -0.1117
RegNAIS -0.0224 -0.0147 -0.0232 -0.0178 0.0781

 

0 Source: Calculated from Survey Data Estimation

Younger producers from the Southeast, not using local auctions, with registered
NAIS premises see greater importance (are more likely to respond very important) for
increasing consumer confidence.

For every ten years a producer’s age increases they are 4.60% less likely to
respond that traceability is very important for increasing consumer confidence.

Producers in the Southeast are 24.21% more likely than producers in the base region to

 

'4 Table 7.7a shows the estimated coefficients and standard errors.

87

respond that traceability is very important for increasing consumer confidence. '5

Producers using local auctions are 11.17% less likely to respond that traceability is very
important for increasing consumer confidence. Producers with their premises registered
in NAIS are 7.81% more likely to respond that traceability is very important for

increasing consumer confidence.

7.1.5 Producer Concerns When Implementing Traceability Systems

Continuing with the second tier of objectives, this thesis evaluated if previously
identified concerns regarding traceability within the beef industry were still of top
apprehension to participation or have producers perceptions shifted. The second core
question for these objectives examined responses to the question: “In designing a
national, individual animal traceability system how concerned are you regarding the
following issues in the US. beef industry.” This question was asked with specific
reference to the following concerns: cost to participating producer, liability to
participating producer, confidentiality of information, reliability of technology, non-
participating firms benefiting, and failure of system to meet stated goals. Appendix 3
(Tables 24-241) provides a summary of producer’s answers to this question.

When it came to concerns of implementation of traceability systems producers
were concerned (responded as either concerned or very concerned) with cost (81.43%),
liability (78.01%), reliability of technology (75.04%), failure of system to meet stated
goals (73.92%), confidentiality of information (71 .42%), and non-participating firms

benefiting (60.45%).

 

'5 Base region consists of the following states in the Northern Crescent region of the US: MN, WI, CT,
ME, MD, MA, NH, NJ, NY, PA, RI, and VT. This region comprises 3.70% of the survey population.

88

Table 7.8 presents the ME estimates for how concerned producers are with the

costs associated with traceability systems.'6

Table 7.8. ME to Concerns Regarding System Cost

 

 

 

ME
Entirely Very
Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0013 0.0010 0.0022 0.0022 -0.0066
Years —0.0086 -0.0065 -0.0151 -0.0147 0.0449
OFeed —0.0103 -0.0078 -0.0182 -0.0176 0.0539
Cows -0.0099 -0.0075 -0.0175 -0.0169 0.0518

 

0 Source: Calculated from Survey Data Estimation

We found younger producers are more likely to be very concerned with the entire
list of issues in designing a national, individual animal traceability system. Producers
with more years of experience in larger operations that produce a larger proportion of
their own feed are more likely to view cost to the participating producer as a main
concern (are more likely to respond very concerned). For every ten years a producer’s
age increases they are 6.60% less likely to respond being very concerned with cost to the
participating producers. For every five years of additional experience producers are
4.49% more likely to respond being very concerned with cost to the participating
producers. Increasing the proportion of feed/forage needs that an operation produces on
their own farm by 25% is associated with a 5.39% increase in odds of being very
concerned with cost of a traceability system. For every level of beef cows that calved in
2007 increase (see question 14 in survey, Appendix 1) producers are 5.18% more likely
to respond to being very concerned with cost.

Table 7.9 presents the ME estimates for how concerned producers are with

 

'6 Table 7.8a shows the estimated coefficients and standard errors.

89

increased liability from traceability systems.'7

Table 7.9. ME to Concerns Regarding System Liability

 

 

 

ME
Entirely Very

Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0014 0.0016 0.0037 0.0022 -0.0089
NP -0.0252 -0.0289 -0.0647 -0.0391 0.1579
SE -0.0413 -0.0474 -0.1062 -0.0642 0.2591
SP -0.03 00 -0.0344 -0.0770 -0.0466 0.1879
Years -0.0080 -0.0091 -0.0204 -0.0124 0.0498
OF eed -0.0057 -0.0065 —0.0147 -0.0089 0.0358
Cows -0.0056 -0.0064 -0.0143 -0.0087 0.0349

 

0 Source: Calculated from Survey Data Estimation

Producers in the Northern Plains, Southeast, and Southern Plains, as opposed to
producers in the Northern Crescent, with more years of experience, larger operations that
produce a large proportion of their own feed revealed heightened concern (are more
likely to respond very concerned) over the liability to the participating producer a
traceability system brings.

For every ten years a producer’s age increases they are 8.90% less likely to
respond being very concerned with liability to the participating producer. Producers in
the Northern Plains, Southeast, and Southern Plains are 15.79%, 25.91%, and 18.79%,
respectively, more likely than producers in the base region to respond to being very
concerned with the liability to the participating producer.18 For every five years of
additional experience producers are 4.98% more likely to respond to being very
concerned with liability. Increasing the proportion of feed/forage needs that an operation
produces on their own farm by 25% is associated with a 3.58% increase in odds of being

very concerned with liability. Furthermore, for every level of beef cows that calved in

 

'7 Table 7.9a shows the estimated coefficients and standard errors.
’8 Base region consists of the following states in the Northern Crescent region of the US: MN, WI, CT,
ME, MD, MA, NH, NJ, NY, PA, RI, and VT. This region comprises 3.70% of the survey population.

90

2007 increase (see survey) producers are 3.49% more likely to respond to being very
concerned with liability to the participating producer.
Table 7.10 presents the ME estimates for how concerned producers are with

confidentiality of information. '9

Table 7.10. ME to Concerns Regarding Confidentiality

 

 

 

ME
Entirely Very
Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0020 0.0013 0.0023 0.0009 -0.0064
Years -0.0204 -0.0135 -0.0239 -0.0090 0.0668
OIncome 0.0076 0.0051 0.0089 0.0034 -0.0250
Labor 0.0121 0.0080 0.0142 0.0054 -0.0397

 

- Source: Calculated from Survey Data Estimation

Producers with more years of experience, whose primary source of income is
from the farm operation, and hire a low proportion of labor are more likely to view
confidentiality of information as a top concern (are more likely to respond very
concerned) in program design. For every ten years a producer’s age increases they are
6.40% less likely to respond being very concerned with confidentiality of information.
For every five years of additional experience producers are 6.68% more likely to respond
to being very concerned with confidentiality of information. Increasing the proportion
off-farm income by 20% is associated with a 2.50% decrease in odds Of being very
concerned with the confidentiality of traceability information. Similarly, increasing the
proportion of hired (non-family) labor on an operation by 25% is associated with a 3.97%
decrease in the probability of being very concerned with confidentiality.

Table 7.1 1 presents the ME estimates for how concerned producers are with the

 

'9 Table 7.10a shows the estimated coefficients and standard errors.

91

reliability oftraceability technology.20

Table 7.11. ME to Concerns Regarding Reliability

 

 

 

ME

Entirely Very
Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0008 0.0008 0.0020 0.0008 -0.0045
NP -0.0364 -0.0361 -0.0893 -0.0377 0.1995
NW -0.0391 -0.0387 -0.0958 -0.0405 0.2141
SE -0.0450 -0.0446 -0.1 105 -0.0467 0.2469
SP -0.0408 -0.0404 -0. 1001 -0.0423 0.2236
Years -0.0081 -0.0080 -0.0198 -0.0084 0.0443
Income -0.0047 -0.0047 -0.01 16 -0.0049 0.0259
OFeed -0.0068 -0.0067 -0.0167 -0.0071 0.0373
Cows -0.0049 -0.0048 ~0.01 19 -0.0050 0.0266

 

0 Source: Calculated from Survey Data Estimation

Operators outside the Northern Crescent and the Corn Belt, with more years of
experience, a higher level of household income, on larger operations that produce a larger
proportion of their own feed are more likely to indicate reliability of technology is a
major concern (are more likely to respond very concerned) in designing a national,
individual animal traceability system. For every ten years a producer’s age increases they
are 4.50% less likely to respond being very concerned with the reliability of traceability
technology. Producers in the Northern Plains, Northwest, Southeast, and Southern Plains
are 19.95%, 21.41%, 24.69%, and 22.36%, respectively, more likely than producers in
the base region to respond to being very concerned with the reliability of traceability
technology.21 For every five years of additional experience producers are 4.43% more
likely to respond to being very concerned with reliability of technology. For every level

$25,000 increase in household income producers are 2.59% more likely to indicate being

 

2° Table 7.11a shows the estimated coefficients and standard errors.
2' Base region consists of the following states in the Northern Crescent region of the US: MN, WI, CT,
ME, MD, MA, NH, NJ, NY, PA, RI, and VT. This region comprises 3.70% of the survey population.

92

very concerned with technology reliability. Increasing the proportion of feed/forage
needs that an operation produces on their own farm by 25% is associated with a 3.73%
increase in odds of being very concerned with reliability. Furthermore, for every level of
beef cows that calved in 2007 increase (see survey question 14) producers are 2.66%
more likely to respond to being very concerned with reliability of the technology of an
implemented traceability system.

Table 7.12 presents the ME estimates for how concerned producers are with non-

participating firms benefiting.22

Table 7.12. ME to Concerns Regarding Non-Participating Firms Benefiting

 

 

 

ME
Entirely Very
Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0013 0.0011 0.0020 -0.0003 -0.0041
Years -0.0118 -0.0098 -0.0175 0.0025 0.0364
OFeed -0.0102 -0.0085 -0.0152 0.0022 0.0316
Cows -0.0071 -0.0059 -0.0106 0.0015 0.0221

 

0 Source: Calculated from Survey Data Estimation

Producers with more years of experience with larger operations that produce a
large proportion of their own feed are more likely to view non-participating firms
benefiting as a key concern (are more likely to respond very concerned). For every ten
years a producer’s age increases they are 4.10% less likely to respond being very
concerned with non-participating firms benefiting from implemented traceability
systems. For every five years of additional experience producers are 3.64% more likely
to respond to being very concerned with non-participating firms benefiting. Increasing
the proportion of feed/forage needs that an operation produces on their own farm by 25%

is associated with a 3.16% increase in odds of being very concerned with non-

 

” Table 7.12a shows the estimated coefficients and standard errors.

93

participating firms benefiting. Furthermore, for every level of beef cows that calved in
2007 increase (see survey question 14) producers are 2.21% more likely to respond to
being very concerned with non-participating firms benefiting.

Table 7.13 presents the ME estimates for how concerned producers are with the

failure of a traceability system to meet stated goals.23

Table 7.13. ME to Concerns Regarding System Failures

 

 

 

ME

Entirely Very
Parameter Unconcemed Unconcemed Neutral Concerned Concerned
Age 0.0008 0.0007 0.0017 0.0006 -0.0037
NP -0.0398 -0.0333 -0.0807 -0.0286 0.1825
NW -0.0396 -0.0332 -0.0804 -0.0285 0.1817
SE -0.0441 -0.03 70 -0.0895 -0.0317 0.2023
SP -0.0500 -0.0419 -0.1014 -0.0359 0.2291
Years -0.0083 -0.0070 -0.0169 -0.0060 0.0383
OFeed -0.0100 -0.0084 -0.0203 -0.0072 0.0459

 

0 Source: Calculated fi'om Survey Data Estimation

Cow-calf producers in the Northern Plains, Northwest, Southeast, and Southern
Plains, as opposed to producers in the Northern Crescent, with more years of experience
in operations that produce a large proportion of their own feed are more likely to be very
concerned with the failure of a traceability system to meet stated goals.

For every ten years a producer’s age increases they are 3.70% less likely to
respond being very concerned with non-participating firms benefiting from implemented
traceability systems. Producers in the Northern Plains, Northwest, Southeast, and
Southern Plains are 18.25%, 18.17%, 20.23%, and 22.91%, respectively, more likely than
producers in the base region24 to respond to being very concerned with the failure of a

system to meet stated goals. For every five years of additional experience producers are

 

2’ Table 7.13a shows the estimated coefficients and standard errors.
2’ Base region consists of the following states in the Northern Crescent region of the US: MN, WI, CT,
ME, MD, MA, NH, NJ, NY, PA, RI, and VT. This region comprises 3.70% of the survey population.

94

3.83% more likely to respond being very concerned with the failure of a traceability
system to meet stated goals. Increasing the proportion of feed/forage needs that an
operation produces on their own farm by 25% is associated with a 4.59% greater
likelihood of producers being very concerned with the failure of a traceability system to

meet stated goals.

7.1.6 Issues in the Beef Industry when Implementing Traceability Systems

In the third and final set of producer perceptions we evaluated the question,
“Please circle your level of agreement with each of the following statements concerning
implementing individual animal traceability systems: Is more cost effective for larger
cow-calf operations, results in more liability for cow-calf producers than cattle owners at
other stages of production, is unnecessary if COOL (Country-of-Origin Labeling) was
implemented nationally, and as a mandated system is exaggerated in need. " Appendix 3
(Tables 25-25d) provides a summary of producer’s answers to this question.

This sample summary reveals that 47.50% of cow-calf producers believe COOL
is more important than a proposed traceability system. Table 7.14 presents the ME

estimates of producers’ perceptions concerning COOL implementation”.

Table 7.14. ME to Implementation of COOL

 

 

 

ME

Strongly Strongly
Parameter Disagree Disagree Neutral Agree Agree
Education 0.01 18 0.011 1 0.0053 -0.0062 -0.0221
Years -0.0151 -0.0143 -0.0068 0.0079 0.0283
OIncome 0.0139 0.0131 0.0063 -0.0073 -0.0260
Auctions -0.03 73 -0.0352 -0.0168 0.0195 0.0697
RegNAIS 0.0468 0.0442 0.021 1 -0.0245 -0.0876

 

0 Source: Calculated from Survey Data Estimation

 

25 Table 7.14a shows the estimated coefficients and standard errors.

95

Cow-calf operators with more years of experience, lower education, with a large
percentage of income from the farm, who use local auctions as a marketing outlet, and do
not have their premises registered with NAIS are more likely to agree that implementing
individual traceability systems is unnecessary if COOL was implemented nationally. For
every ten years a producer’s age increases they are 2.21% less likely to agree that
traceability is unneeded if COOL was implemented. For every five years of additional
experience, producers are 2.83% more likely to agree that traceability is unneeded if
COOL was implemented. Every education level increase (see survey question 5) causes
producers to be 2.21% less likely to agree that COOL could replace traceability.
Producers who use local auctions to market their cattle are 6.97% more likely, while
producers with registered premises in NAIS are 8.76% less likely to agree that
implementing individual traceability systems is unnecessary if COOL was implemented
nationally. Results concerning the remaining statements in this question can be found in

Appendix 4.

7.2 Choice Experiment Analysis

The following set of estimates reveals the results from the MN L, RPL, and LCM
(3-segment) model estimations. The RPL and LCM models were both preferred (via
Log-likelihood ratio (LR) tests) to the MNL model. This suggests significant preference

heterogeneity exists among our sample of producers.

7.2.1 Multinomial Logit Model

Table 7.15 presents the MNL point estimates as well as standard errors.

96

Table 7.15. Multinomial Logit Estimates

 

 

Parameter Coefficient Standard Error
PD 0.0340*** 0.0073
No Trace -0.0485 0.4532
Advanced Trace -0.2518 0.4472
Pvtl -0.0159 0.1043
PvtNI 0.1013 0.1032
PP 0.6369*** 0.2345
PG 0.5227” 0.2309
HR -0.1135 0.2456
AVPP -0.1508 0.1577
AVPG -0. 1945 0.2452
AVHR 0.3881** 0.1682
PPPG -0.2067 0.2430
PPHR -0.2383 0.1584
PG & HR 0.0029 0.2275
AVPPPG -0.0508 0.2331
AVPPHR 0.4818*** 0.1694
PPPGHR -0.2124 0.1621
AVPPPGHR -0.9330*** 0.2416
Adv Trace * Cows 0.0620 0.0406
No Trace * Cows 0.0973*** 0.0440
Adv Trace "‘ Auctions -0.1186 0.1232
No Trace "‘ Auctions 0.4527*** 0.1343
Adv Trace * Member 0.1011“ 0.0496
No Trace * Member -0.1980*** 0.0538
Adv Trace * Years -0.1099*** 0.0390
No Trace * Years 0.0129 0.0422
Adv Trace * IndvID 0.9874* 0.2583
No Trace * IndvID 0.0965 0.2130
Adv Trace * GrpID 0.0345 0.1595
No Trace * GrpID -0.2513 0.1689
Adv Trace * NotID 0.7568* 0.3886
No Trace * NotID 1.0421*** 0.3543
Adv Trace * CB -0.3336 0.3091
No Trace * CB -0.2952 0.3370
Adv Trace * NP -0.1149 0.3051
No Trace * NP -0.0363 0.3353
Adv Trace * NW -0.4340 0.3102
No Trace * NW 0.0807 0.3398
Adv Trace * SE -0.8857*** 0.3133
No Trace * SE -0.5664* 0.3412
Adv Trace * SP -0.9271*** 0.3210
NO Trace * SP -0.2818 0.3457

 

0 Source: Calculated fi'om Survey Data Estimation
o *** 1% significance level, ** 5% significance level, * 10% significance level

97

The MNL model results (Table 7.15) Show that, when comparing alternative
traceability programs, cow-calf producers are sensitive to price adjustments. This
suggests voluntary participation in traceability programs deemed undesirable to the
representative cow—calf producer may be accomplished by offering premiums for
participation, it also suggests participation may be enhanced by markets penalizing non-
participants. As previously noted, this evaluation of Advanced Traceability systems
considered multiple pieces (and combinations) of required information to be provided.
The MN L model suggests producers would prefer to provide production practice (PP) or
performance/ genetic (PG) information rather than provide age verification (AV).
However, for programs characterized with AV as a requirement, producers appear to
prefer providing both AV and health certification/vaccination records (HR) instead of
solely AV. Similarly, producers prefer to provide the combination of AV, PP, and HR
information, rather than solely provide AV. This suggests producers may see value in
providing this additional information once AV is required. However, producers do have
a strong preference against programs requiring all four evaluated information components
of AV, PP, PG, and HR. This evaluation allowed us to rank the preferred information
content components of Advanced Traceability systems. Implied rankings are as
followed: (1) AV, PP, and HR, (2) PG, (3) PP, (4) AV and HR, (5-13) AV and the
insignificant (relative to AV) combinations, and (14) AV, PP, PG, and HR.

The inclusion of demographics (via interaction terms) in the MN L model allowed
for further examination of the types of producers valuing NAIS, Advanced Traceability,
and No Traceability options differently. Producers who are active in industry

associations, have less years of experience, use individual or no/other animal

98

identification, and do not operate in the Southern Plains are more likely to prefer
Advanced Traceability systems over NAIS. Alternatively, producers who have larger
operations, use local auctions to market their cattle, are less active in industry
associations, use no/other animal identification, and are not from the Southeast are more
likely to prefer No Traceability to NAIS.

Individual coefficients are generally of limited value with more insights being
ascertained from combinations of multiple coefficients. Of particular interest are
estimates of producers’ willingness-to-accept (WTA) estimates. Table 7.16 provides
WTA estimates for the MNL model. The No Traceability (N 0 Trace) and Advanced
Traceability (Adv Trace) WTA values were calculated at the mean values of each
producer demographic variable (Cows, Auctions, Member, Years, IndvID, GrpID, NotID,

CB, NP, NW, SE, and SP) included in the model.

Table 7.16. Multinomial Logit Willingness-to-Accept

 

 

Parameter WTA 95% Confidence Interval

No Trace (vs. NAIS) $2.17 [$14.28, -$13.36]
Adv Trace (vs. NAIS) -$l.78 [$6.76, -$7.70]
PvtI (vs. Gov’t) -$0.93 [$12.19, -$l4.34]
PvtNI (vs. Gov’t) $5.96 [$19.56, -$6.27]
PP (vs. AV) $37.43 [$70.30, $11.28]
PG (vs. AV) $30.72 [$65.24, $5.42]
HR (vs. AV) -$6.67 [$23.67,-$37.97]
AVPP (vs. AV) -$8.86 [$10.47, -$31.32]
AVPG (vs. AV) -$11.43 [$22.28, -$42.24]
AVHR (vs. AV) $22.81 [$47.75, $3.78]
PPPG (vs. AV) -$12.15 [$16.01, -$43.48]
PPHR (vs. AV) -$14.01 [$6.25, -$36.58]
PGHR (vs. AV) $0.17 [$29.32, -$28.43]
AVPPPG (vs. AV) -$2.98 [$28.04, -$34.48]
AVPPHR (vs. AV) $28.32 [$51.45, $10.13]
PPPGHR (vs. AV) -$12.49 [$4.87, -$35.96]
AVPPPGHR (vs. AV) -$54.84 [-$30.25, -$91.15]

 

0 Source: Calculated from Survey Data Estimation
0 All presented values are in $/animal.

99

It is worth noting that the results of the MNL model imply the representative
producer is indifferent between the three evaluated programs (NAIS, Advanced
Traceability, and No Traceability) as well as to the managing entity of the evaluated
programs (insignificant WTA estimates). However, producers appear to be sensitive to
the information required for participation. In comparing two traceability systems
requiring only one piece of additional information, producers must be paid $3 7.43/animal
in a system seeking AV to make them indifferent between providing AV and providing
PP information in an alternative system. Similarly, producers will accept discounts
(relative to alternative traceability systems) up to $30.72/animal before they will
participate in a system requiring AV rather than a system solely requiring PG
information. However, as noted above, in evaluating multiple traceability systems that
each require the provision of AV, producers appear to prefer providing additional
information. In particular, table 7.16 suggests that a traceability system requiring both
AV and HR or requiring in combination AV, PP, and HR are preferred to those solely
requiring AV. The estimates of $22.81/animal and $28.32/animal imply producers would
actually accept discounts of these two amounts for providing these information
combinations, rather than simply providing AV.

While this result may seem counter-intuitive, a potential explanation would be
producers are anticipating some reward (beyond immediate price/head adjustments) for
providing this information and hence are readily willing to provide it in systems
characterized by age verification requirements. That is, AV may be a “threshold
attribute” that once required results in other attributes (PP and HR) being easily provided

as well. However, producers maintain a preference to not be required to provide too

100

much information. This is evident from the estimate that producers would accept a
discount of $54.84/animal before they would provide all four information components
(AV, PP, PG, and HR) rather than simply provide AV.

AS previously noted, standard MNL models may be limiting because they do not
allow for random taste variation, unrestricted substitution patterns, and correlation in
unobserved factors over time. Therefore, additional models, including RPL and LCM

models, were estimated and results shown below.

7.2.2 Random Parameter Logit Model
As previously mention a LR test lead to rejection of the MNL model for the RPL

model. Table 7.17 presents point estimates and standard errors for the RPL model.

Table 7.17. Random Parameter Logit Estimates

 

 

Parameter Coefficient Standard Error
Random parameters in utility functions
No Trace -1.0278 2.3343
Adv Trace -l .8904 1.6992
Nonrandom parameters in utility functions
PD 0.1226*** 0.0174
PvtI 0.2909 0.2016
PvtNI -0.0363 0.2322
PP 2.0239*** 0.5233
PG l.3677*** 0.5027
HR -1.0858* 0.6112
AVPP 0.0821 0.3799
AVPG -1.5800** 0.6474
AVHR 0.7959" 0.3504
PPPG -l.0351* 0.5525
PPHR —0.5817 0.3606
PGHR 0.1060 0.5631
AVPPPG 0.2580 0.5641
AVPPHR 1.6073 * * * 0.5066
PPPGHR -0.2458 0.3308
AVPPPGHR -1 .8135*** 0.5322
Adv Trace * Cows 0.2581 0.1990
No Trace * Cows 0.4687* 0.2666

101

Table 7.17 (cont’d)

 

 

Parameter Coefficient Standard Error
Adv Trace * Auctions -0.0833 0.5649
No Trace * Auctions 1.5573* 0.8051
Adv Trace * Member 0.0506 0.2438
No Trace * Member -0.8l79** 0.3298
Adv Trace * Years -0.3062 0.1883
No Trace * Years -0.0056 0.2693
Adv Trace * IndvID 2.7739“ 1.0833
No Trace * IndvID 0.0494 1.4130
Adv Trace * GrpID 0.2023 0.6674
No Trace * GrpID -0.5046 0.9526
Adv Trace * NotID 2.4064 1.6834
No Trace * NotID 3.2783 2.0482
Diagonal values in Cholesky matrix, L.
No Trace 6.6582*** 0.6847
Adv Trace 3.5514*** 0.3106
Below diagonal values in L matrix.
Adv Trace: No Trace 2.66525*** 0.5517
Standard deviations of parameter distributions
Stand. Dev. of No Trace 6.6582*** 0.6847
Stand. Dev. of Adv Trace 4.4402*** 0.4233

 

0 Source: Calculated from Survey Data Estimation
0 *** 1% significance level, ** 5% Significance level, * 10% Significance level

The RPL model results (Table 7.17) show that, when comparing alternative
traceability programs, cow-calf producers are sensitive to price adjustments. As noted in
the previous model, these RPL model results suggest that participation in traceability
programs deemed undesirable to the representative cow-calf producer may be
accomplished by offering premiums for participation or participation may be influenced
by markets penalizing non-participation. The evaluation of Advanced Traceability
systems through the RPL model considered the same multiple pieces (and combinations)
of required information to be provided to producers as was the case in the MNL model.
As in the MNL model, the RPL model suggests producers would prefer to provide PP or
PG information rather than provide AV. On the other hand, the RPL model implies that

producers would prefer to provide HR as opposed to providing AV. For programs

102

characterized with AV, as a requirement, producers appear to prefer providing both AV
and PP, as well as, AV and HR instead of solely AV. Similarly, producers prefer to
provide the combination of AV, PP, and HR information, rather than solely provide AV.
As previously explained, this suggests producers may see value in providing this
additional information once AV is required. However, producers do have a strong
preference against programs requiring all four evaluated information components of AV,
PP, PG, and HR. Further evaluation reveals that producers would prefer to provide AV
individually to AV and PG or PP and PG. An implied ranking of the preferred
information content components of advanced traceability systems was: (1) PP, (2) AV,
PP, and HR, (3) PG, (4) AV and HR, (5) AV and PP, (6 -10) AV and the insignificant
(relative to AV) combinations, and (l 1) PP and PG, (12) HR, (1 3) AV and PG, and (14)
AV, PP, PG, and HR.

The inclusion of demographics (via interaction terms) in the RPL model allowed
for further examination of the types of producers valuing NAIS, Advanced Traceability,
and No Traceability options differently. Producers who use individual animal
identification are more likely to prefer Advanced Traceability systems over NAIS.
Producers who have larger Operations, use auctions, and are less active in industry
associations are more likely to prefer No Traceability to NAIS.

To further evaluate preference heterogeneity in the RPL model the estimated
Cholesky matrix was examined. The diagnol values of the Cholesky matrix represent the
true value of variance for each random parameter once the cross-correlated parameters
terms have be uncofounded (Hensher, Rose, and Green, 2006). Both random parameters,

No Trace and Adv Trace, were estimated to have statistically significant diagnol

103

Cholesky elements. This provides evidence of preference heterogeneity persisting for
each traceability system, even after incorporating producer characteristics.

Examination of the off-diagnol elements (Adv Trace: No Trace) of the Cholesky
matrix reveals a statistically significant (0.001) estimate. This suggests a significant
cross-correlation among the random parameter estimates would have been inappropriate]
confused within standard deviation estimates of each random parameter without
Cholesky matrix decomposition and evaluation (Tonsor et al., 2008). Evaluation of the
correlation term reveals the No Trace and Adv Trace variables to generally be positively
correlated within the population. This suggests Advanced Traceability and No
Traceability are closer substitutes than suggested by the non-stochastic portion of the
model (Alfines, 2004).

Interpretation of individual coefficients must be made with caution and is
generally discouraged in random utility models (Scarpa and Del Giudice, 2004). Caution
is particularily warranted in these models given differences in scales and the number of
interaction terms between producer characteristics and traceability system attributes

(Tonsor et al., 2008). Table 7.18 provides WTA estimates for the RPL model.

104

Table 7.18. Random Parameter Logit Willingness-to-Accept

 

 

Parameter WTA 95% Confidence Interval

No Trace (vs. NAIS) $0.51 [$11.90, -$12.16]
Adv Trace (vs. NAIS) $1.10 [$7.98, -$5.57]
PvtI (vs. Gov’t) $4.75 [$11.97, ~$1.86]
PvtNI (vs. Gov’t) —$0.59 [$7.23, -$7.91]
PP (vs. AV) $33.03 [$51.58, $16.29]
PG (vs. AV) $22.32 [$39.86, $6.11]
HR (vs. AV) -$17.72 [$3.19, -$38.46]
AVPP (vs. AV) $1.34 [$13.19, -$10.91]
AVPG (vs. AV) -$25.79 {-$4.21, -$50.67]
AVHR (vs. AV) $12.99 [$24.82, $1.40]
PPPG (vs. AV) -$16.89 [$0.46, -$35.46]
PPHR (vs. AV) -$9.49 [42.45, -$20.93]
PGHR (vs. AV) $1.73 [$21.28, -$18.61]
AVPPPG (vs. AV) $4.21 [$21.70, -$15.17]
AVPPHR (vs. AV) $26.23 [$42.54, $9.49]
PPPGHR (vs. AV) -$4.01 [$6.25, -$15.34]
AVPPPGHR (vs. AV) -$29.60 [-$11.96, -$48.54]

 

0 Source: Calculated from Survey Data Estimation
o All presented values are in $/animal.

As was the case in the MNL model, these results imply producers are indifferent
between the three evaluated programs (NAIS, Advanced Traceability, and No
Traceability) as well as to the managing entity of the evaluated programs. However,
producers appear to be sensitive to the information required for participation. Producers
must be compensated $33.03/animal in a system seeking AV to make them indifferent to
participating in an alternative system solely requiring PP information. Similarly,
producers will accept discounts up to $22.32/animal before they will provide AV rather
than PG information. In evaluating multiple traceability systems that each require the
provision of AV, producers appear to prefer providing some combinations of additional
information. In particular, Table 7.18 suggests that a traceability system requiring both
AV and HR or requiring in combination AV, PP, and HR are preferred to those solely

requiring age verification. The estimates of $12.99/animal and $26.23/animal imply

105

producers would actually accept discounts of these two amounts for providing these
information combinations, rather than simply providing AV. As explained previously,
producers could be anticipating some reward for providing this information and hence are
readily willing to provide it in systems characterized by age verification requirements.
However, producers maintain a preference to not be required to provide too much
information. This is evident from the estimate that producers would accept a discount of
$25.79/animal before they would provide AV and PG rather than provide AV solely.
Similarly, producers would accept a discount of $29.60/animal before they would provide
all four information components (AV, PP, PG, and HR) rather than simply provide AV.
Although RPL models do account for heterogeneity, a LCM strategy may be more
suited for capturing this heterogeneity as it assumes that producers can be separated into
different segments such that producers within each segment have homogeneous
preferences, but that preferences vary across producers from alternative segments.
Thereby identifing marketing segments of producers likely to support alternative
traceability systems. Furthermore, LCM models have been found to be better suited for
identifying individuals with heterogeneous preferences thanRPL models (Ouma et al.,

2007)

7.2.3 Latent Classification Model

The LCM model was determined to be preferred to the RPL model because the
LCM model accounted for producer heterogeneity and allowed for the identification of
marketing segments. Furthermore, utilizing a LR test determined the LCM model was

preferred to the MNL model. The segmentation of producers into marketing segments

106

will help better identify how traceability systems should be designed and promoted in
order to improve voluntary participation rate.

Following Boxall and Adamowicz (2002) allowed for identification of the number
of segments minimizing the Bayesian Information Criterion (BIC), with three segments
being preferable.26 This LCM 3-Segment model evaluates producer’s preferences across
three very distinct segments. For the purposes of interpretation segment one is named the
Advanced Traceability Preferring segment (37.1% of producers), segment two the No
Traceability Preferring segment (27.4%), and segment three the NAIS Traceability
Preferring segment (35.5% of producers). These segment names are only used for ease
in interpretation, with each segment being comprised of many more distinct
characteristics that will be discussed in the following interpretation. Table 7.19 shows

the coefficient and standard errors for the 3-segment LCM Model.

 

26 Alternative models, for instance containing different covariates, in some cases produced models where 4-
segments were preferred. However, in certain cases these 4-Segment models had convergence issues,
therefore a 3-segment model was selected.

107

Table 7.19. Latent Class Model Estimates

 

 

 

 

 

 

 

Advanced
Traceability No Traceability NAIS Traceability
Preferring Preferring Preferring

Std Std

Parameter Coefficient Err Coefficient Std Err Coefficient Err
PD 0.0707*** 0.0215 0.0667“ 0.0327 0.0887*** 0.0117
No Trace 0.0704 0.3509 4.6459*** 0.5991 -2.2121*** 0.1963
Adv Trace 1.8530*** 0.3621 0.3488 0.6591 -3.0429*** 0.2233
Pvtl -0.1030 0.2455 1.0780*** 0.4038 0.4159*** 0.1575
PvtNI 0.5003 0.3196 -0.1010 0.3956 0.2538 0.1643
PP 0.4558 0.7943 3.1588*** 0.7521 1.2545*** 0.3185
PG ~0.6730 0.7102 2.6228*** 0.6179 1.1112*** 0.2962
HR -0.3344 0.7238 0.1829 0.8623 -1.8159*** 0.5961
AVPP 0.7713 0.6458 10717“ 0.5877 -0.1700 0.2856
AVPG -1.0177** 0.4678 -3.9381 3.9887 -1.6913*** 0.6392
AVHR -0.4046 0.4350 -0.0289 0.8684 0.3256 0.2730
PPPG -1.1796*** 0.4438 -1.8622 3.9656 1.2823*** 0.4321
PPHR 0.0836 0.6106 -0.2602 0.7805 -0.4239 0.3019
PGHR 2.1897 1.3511 -2.2191 4.0670 0.6836 0.4732

----- Demographics------
THETA(1) in Class THETA (2) in Class
Model model Base

Std Std Std

Coefficient Err Coefficient Err Coefficient Err

Constant -0.6557 0.7791 -0.1864 0.7148 N/A N/A

Cows 0.0553 0.0815 0.1178 0.0884 N/A N/A

Auctions -0.1560 0.2424 0.4719* 0.2666 N/A N/A

Member 0.0814 0.0985 -0.2628** 0.1107 N/A N/A

Years -0.0787 0.0731 -0.0831 0.0731 N/A N/A

lndvID 1.0116* 0.5550 0.0740 0.4334 N/A N/A

GrpID -0.0346 0.2923 -0.06 l 6 0.3051 N/A N/A

NotID 1.0265 0.8799 1.2244 0.7565 N/A N/A

 

0 Source: Calculated from Survey Data Estimation
0 *** 1% Significance level, ** 5% Significance level, * 10% significance level

108

The LCM results (Table 7.19) Show that, when comparing alternative traceability
programs, cow-calf producers in each segment are sensitive to price adjustments.
Suggesting, consistent with the MNL and RPL model results, that participation in
traceability programs deemed undesirable to the representative cow-calf producer may be
accomplished by offering premiums for participation or participation may be influenced
by markets penalizing non-participation. The evaluation of Advanced Traceability
systems through the LCM considered the single pieces and pairs of required information
to be provided to producers.27

In evaluating the Advanced Traceability Preferring class (segment 1), the LCM
suggests producers would prefer Advanced Traceability systems. Results show that
producers would prefer to solely provide AV over providing AV and PG information or
providing PP and PG information. Producers in the No Traceability Preferring class
(segment 2) prefer No Traceability as the name suggests, as well as prefer a system that
has a private-Industry managing entity. These producers would prefer to provide PP or
PG information to providing AV solely. However, for programs characterized with AV,
as a requirement, producers appear to prefer providing both AV and PP information
instead of individually AV. The final class, NAIS Traceability Preferring (segment 3),
producers prefer NAIS Traceability to No Traceability and Advanced Traceability. Note
that these producers also prefer No Traceability to Advanced Traceability. Within a
traceability system they would prefer a system with a private-industry managing entity.
For this NAIS Preferring class the LCM suggests producers would prefer to provide PP or

PG information or PP and PG information rather than provide just AV. On the other

 

27 Due to convergence issues with the LCM model only single and pairs of additional information were
included, as opposed to the full set of additional information combinations that were included in the MNL
and RPL models.

109

hand, this model shows that producers would prefer to provide AV as Opposed to
providing HR. As shown with previous models producers do have a preference against
programs requiring too much information, as producers would rather provide AV than
AV and PG information.

Inclusion of demographics in the LCM allows further examination of the types of
producers valuing NAIS, Advanced Traceability, and No Traceability options differently.
Producers using individual animal identification are more likely to be in segment I and
prefer Advanced Traceability to NAIS Traceability. Producers who are less active
amongst industry associations and use local auctions are more likely to be in segment 2
and prefer No Traceability to NAIS Traceability.

As previously mentioned, individual coefficients are generally of limited value
with more insights being ascertained from combinations of multiple coefficients. Of
particular interest are estimates of producers’ WTA estimates. Table 7.20 provides the

WTA estimates for the LCM model.

110

Table 7.20. Latent Class Model Willingness-to-Accept

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Advanced Traceability No Traceability NAIS Traceability
Preferring Preferring Preferring

Parameter WTA WTA WTA
No Trace -$4.63 $55.00 -$32.49
@s. NAIS) [$11.40, -$28.43] [$259.07, $21.34] [-$21.59, -$46.55L
Adv Trace $26.22 $5.23 -$34.30
(vs. NAIS) [$77.13, $11.69] [$60.07, -$18.06] [-$27.23, -$43.42j
Pvtl -$2.91 $32.32 $9.38
(vs. Gov’t) [$11.69, -$21.35] [$119.71, $10.04] [$17.88, $2.67]
PvtNI $14.16 -$3.03 $5.72
(vs. Gov’t) [$40.20, -$4.36] [$48.07, -$35.53] [$13.17, -$1.69]
PP $12.90 $94.71 $28.28
(vs. AV) [$69.55, -$36.19] [$407.71, $44.57] [$44.45, $14.15]
PG -$l9.05 $78.64 $25.05
(vs. AV) [$24.89, -$65.98] [$428.55, $23.03] [$43.24, $11.93]
HR -$9.46 $5.48 -$40.94
(vs. AV) [$49.39, -$45.56] [$212.86, -$33.28L [-$17.23, -$69.04]
AVPP $21.83 $32.13 -$3.83
(vs. AV) [$76.76, -$l6.47] [$153.90, -$8.44] [$9.83, -$15.34]
AVPG -$28.80 -$118.07 -$38.13
(vs. AV) [-$2.23, -$82.74] [$193.71, -$739.35] [-$10.58, -$73.26]
AVHR -$l 1.45 -$0.87 $7.34
(vs. AV) [$12.97, -49.45] [$157.23, -$43.71] [$23.82, -$4.68]
PPPG -$33.39 -$55.83 $28.91
(vs. AV) [~$6.41, -$108.96] [$293.13, -$625.22] [$47.41, $9.90]
PPHR $2.37 ~$7.80 -$9.56
(vs. AV) [$44.80, -$36.97] [$70.48, -$1 1 1.33] [$4.46, -$24.01]
PGHR $61.98 -$66.54 $15.41
(vs. AV) [$184.17, -$12.72] [$272.99, -$692.21] [$37.10, -$5.25]

 

0 Source: Calculated from Survey Data Estimation
o All presented values are in $/animal.

111

The WTA estimates are explained class by class. Beginning with the Advanced
Traceability Preferring class, producers in an Advanced Traceability system must be
compensated $26.22/animal to participate in NAIS and be indifferent to Advanced
Traceability without a premium. However, insignificance of the No Traceability WTA
estimate suggests producers in this class would not require significant compensation to
participate in NAIS rather than in a No Traceability system. Furthermore, producers will
accept discounts up to $28.80/animal before they will provide AV and PG information
rather than solely provide AV. Similarly, these producers will accept discount up to
$33.39/animal before they provide PP and PG information as opposed to only providing
AV.

Producers in the No Traceability Preferring class, as they prefer No Traceability,
would need a premium of $55.00/animal to be indifferent between No Traceability and
NAIS. These producers prefer traceability system managing entities to be private and
within the industry and would need a premium of $32.32/animal to be indifferent to a
traceability system that is managed by the government. Furthermore, producers in this
class would require a $94.71/animal and $78.64/animal premium to provide AV as
opposed to PP or PG information, respectively.

The final set of producers, the NAIS Preferring class, would accept a discount of
up to $32.49/animal before they are indifferent between NAIS and No Traceability and
would accept a discount of up to $34.30/animal before they are indifferent between NAIS
and Advanced Traceability. While revealing a preference for NAIS, these producers
would prefer the managing entity to be private and within the industry. They would

require a premium of $9.38/animal premium to be indifferent between private-industry

112

and government as a managing entity. Producers in this class would require a
$28.28/animal and $25.05/animal premium to provide AV as opposed to PP or PG
information, respectively. However, to provide HR over AV these same producers would
need a premium of $40.94/animal. Considering pairs of information attributes, this class
would need a premium of $38.13/animal to provide AV and PG information as opposed
to just AV. In contrast, the NAIS preferring producers would need a premium of

$28.91/animal to provide AV over PP and PG information in combination.

7.3 Producer Welfare Effects

As discussed in the previous chapter (section 6.7), producer welfare effects,
reflecting the amount necessary to make a producer equally well-off in two alternative
situations that cow-calf producers may face in the future were estimated. More
specifically, this thesis considered situations in which producers face all three traceability
options (i.e., NAIS, Advanced, and No Traceability) versus situations in which the No
Traceability option is removed or both No Traceability and Advanced Traceability are
removed. The latter case would be a situation of mandatory NAIS Traceability. These
remain realistic possibilities, making them valid scenarios to consider. Table 7.21

presents the welfare effect estimates for the MNL, RPL, and LCM models.

113

42553 E 0.8 82? @3385 =< o
scams—Em 8mm xogm an 932330 682% o

 

Gwen .meai 43.3mm- .533

SSS- .382

”2.9%- .Eei

5.43..- 3.82 £388: 32 names:

 

 

 

Nmém- 3.3%. 3.43.. cm. mm- 3.4mm-
Eng- 523 5.3a- .3533 5.3- £22 82%. .232 Lemma- «2 _Z 8&0
m OS. 3. mm. omdm- om. _ m- 05.4 g m- LAD—Enough oZ mo REEDM
chomoE warn—£8; choon
bassoon; 5:588; 0 Z b=5moomcl
m~<Z 825654 Dortm—
EUA and 47:2

 

385 2£63 42 can...

114

Welfare estimates utilizing the MNL model show that a Situation, in which the No
Traceability option is prohibited, would cause a welfare loss of $14.76 per animal.
Along these same lines, imposing mandatory NAIS Traceability suggests a welfare loss of
$34.51/animal. These valuations are estimates of the amount the representative producer
would require making them equally well-off as they were given the option of selecting
any of the three traceability alternatives.

The RPL model suggests removal of the No Traceability option results in an
estimated welfare loss of $1 .59/animal. While imposing mandatory NAIS Traceability
suggests a welfare loss of $3.84/animal.

Utilizing our LCM model, the welfare estimates involved calculating the effects
for the three distinct producer segments. The Advanced Traceability Preferring
producers would have a welfare loss of $2.20/animal if the No Traceability option was
removed and face a welfare loss of $34.07/animal if NAIS Traceability was the only
option available. The No Traceability Preferring producers are far more negatively
impacted by reduced choice sets. These producers face a welfare loss of $73.61/animal
for the removal of No Traceability and a welfare loss of $84.41/animal for a situation in
which only mandatory NAIS is available. Calculations suggest that the NAIS Prefizrring
producers would have a welfare loss of $2.3 5/animal if the No Traceability option was
removed and a welfare loss of $4.52/animal if the No Traceability and Advanced
Traceability options were removed.

Comparison of welfare effect estimates across the three models shows notable
differences. For example welfare estimates for the MNL model are notably higher than

the same estimates in the RPL model. While the segmentation of the LCM model show

115

varying degrees of the welfare effects, as the Advanced Traceability Preferring class has
a relatively small estimate when only removing the No Traceability option, but becomes
considerably larger once the No and Advanced Traceability options are both removed.
Within the NAIS Prefierring class, producer’s welfare effects are relatively small for both
situations. Conversely, producers within the No Traceability Preferring class experience
relatively large welfare effects for both situations that exceed those of the other two LCM
model segments and those suggested by the MNL and RPL models. Collectively, these
results provide further evidence for the importance of model specification and

consideration of producer preference heterogeneity.

7.4 Implications

The above mentioned WTA estimates, and associated inferences regarding
producer preferences, provide a clearer understanding of how traceability systems should
be designed and promoted in order to improve voluntary participation rates. To
summarize the implications of the WTA findings, two specific cases that may benefit
from this research are discussed. The first case would be a scenario of developing a
voluntary traceability system that is fairly similar to NAIS in that there are limited
“additional information pieces” required for producer participation. The purpose of a
heavily structured (e. g., little to no variability in the trade-off between participation
requirements and per animal compensation) traceability system could be to enhance the
entire industry’s ability to quickly identify and contact animal premises, promote animal

identification, and develop animal movement and tracing capabilities. These goals would

116

be consistent with those of animal health management and response objectives held by
USDA in developing and promoting NAIS.

Alternatively, more advanced systems that enable producers or firms further down
the supply chain to use traceability information could be designed as marketing tools.
For instance, a traceability system could be designed to capture the production practice
information most desired by a select consumer group. Provision of this information
would require cooperation and compensation of all players in the supply chain, with
benefits reflecting the perceived final market value of the information to consumers. For
a producer this traceability information would give added value to the cattle they maket,
in exchange for their willing participation. While for a retailer this traceability
information provides value as consumers may be willing to pay for these added attributes.

The discussion of implications will be focused on the three traceability preferring
segments as identified by the Latent Class Model. First, let’s consider the implications of
selecting different managing entities of a traceability system on producer participation.
Our results suggest developers of a system who are primarily interested in achieving
participation rate goals (e. g., as opposed to alternative national herd health goals held by
USDA) can expect systems with managing entities being private and within the industry
to enhance participation by producers. Furthermore, given that the structure of NAIS has
the managing entity being government; these results suggest that producer participation
may have been higher with private management. However, care should be taken in this
assessment as the USDA, in managing NAIS, certainly has objectives beyond simply

maximizing voluntary participation rates (e. g., animal herd health surveillance and

117

response ability enhancement. Nonetheless, the results indicate a trade-off of
participation rate and managing entity does exist for cow-calf producers.

In designing a private traceability system, altering the composition of the
additional information required for participation is found to significantly impact
voluntary participation interest. Advanced systems that only require production practice
information or performance/genetic information, as opposed to age verification, would be
supported within the No Traceability Preferring Class and the NAIS Traceability
Preferring class. Producers in the No Traceability Preferring class would require a
$94.71/animal and $78.64/animal premium to provide age verification as Opposed to
production practice information and performance/ genetic information, respectively. As
such, the cost of requiring producers possessing these preferences (estimated to be 27.4%
of producers) to provide age verification likely outweighs the benefits of increasing the
participation in such a system. As such, traceability system designers need to carefully
outweigh the anticipated benefits of this age verification information (e. g., enhanced
value of marketing verified younger animals) with the compensation that may be required
to obtain participation of producers who may belong to the No Traceability Preferring
class.

Conversely, the estimated premiums for producers in the Advanced Traceability
Preferring class to provide age verification as opposed to production practice,
performance/ genetic information, and heath records, respectively, are not different from
$0. This suggests that small premiums could be paid to easily (and more affordably)
entice producers to increase participation rate. Combined, this suggests that private

traceability systems requiring age verification should first seek out producers belonging

118

to the Advanced Traceability Preferring segment (e. g., those known to use individual
identification methods) and initially avoid expending resources targeting those in th No
Traceability Preferring (e. g., those known to typically market through auctions) class.

Producers in the Advanced Traceability Preferring and No Traceability
Preferring segments are found to be indifferent to systems requiring health verification
records and systems requiring age verification. Conversely, producers in the NAIS
Preferring group hold a preference for providing age verification. Combined, this
suggests that for systems considering verification of animal health throughout the supply
chain to be a higher priority than age verification, target marketing should initially avoid
producers in the NAIS Preferring group.

As a general statement, producers reduce participation in traceability systems as
additional information requirements are added. Furthermore, considering the
implications of requiring pairs of additional information becomes a challenge because
different values may be placed on pairs of attributes (e.g., WTA age verification +
production practice information #5 WTA age verification + WTA production practice
information). Therefore target marketing of systems that are characterized by the pairs of
additional information to provide should be based on the benefits and costs of each
proposed system because results indicate a trade-off of participation rate and additional
information does exist for cow-calf producers.

The above mentioned welfare effect estimates, and associated inferences
regarding implications of producer’s realizing a reduced set of voluntary system options

(e.g., being forced to participate in either Advanced Traceability and/or NAIS

119

Traceability), provides an improved understanding of how requiring traceability or
imposing mandatory NAIS affects producers differently.

As expected, results show that producers in the Advanced Traceability Preferring
segment would have the smallest negative welfare effect if traceability (e. g., Advanced
Traceability or NAIS Traceability) became mandatory. However, if NAIS were to
become mandatory requiring all cow-calf producers to participate, producers in the
Advanced Traceability Preferring segment would be considerably more negatively
affected. This shows that forcing producers to participate in a rather structured, national
traceability system (e. g., NAIS) when they prefer private, advanced traceability system
will likely decrease support.

Producers in the NAIS Preferring segment would experience a relatively small
welfare effect of imposing mandatory traceability, either Advanced Traceability or NAIS
Traceability. This is likely do to their ability to still choose NAIS Traceability which they
prefer. When imposing mandatory NAIS Traceability on these producers there negative
welfare impact becomes slightly larger. This is a result of restricting their choice set.

As one would expect, the results show that imposing mandatory traceability or
mandatory NAIS on producers who prefer No Traceability will have the largest welfare
effect. The implications of these finding require further discussion because producers in
this segment have a strong preference for No Traceability and would be the type of
producers most adversely affected by mandated traceability. The welfare effect estimates
of $73.61/animal and $84.41/animal for removing the No Traceability option and
imposing mandatory traceabily, respectively, may be large enough that increasing support

for traceabilty or mandatory traceability would not be feasible in this segment. Stated

120

differently, these large estimates suggest that these producers will likely not be willing to
voluntarily participate in any traceability system. A prerequisite of their participation
may well be a market characterized by mandated traceability, either by the government of

the private sector that simply chooses to no longer accept “un-traced animals.”

121

Table 7.1a. Producers’ Forecasts

 

Future NAIS Participation

 

 

Future NAIS Participation and RF ID Use
Standard Standard

Parameter Estimate Error Estimate Error
Intercept 42.1771 *** 6.0054 26.1481*** 9.0690
Age 0.0675 0.0930 0.3729*** 0.1068
CB N/A N/A -0.7427 4.7071
NP N/A N/A -1 .9082 4.8144
NW N/A N/A -5.5572 5.0542
SE N/A N/A -0.0020 4.8765
SP N/A N/A -5.8327 5.1530
Member 1.2363 0.7724 0.2839 0.8949
Education N/A N/A -0. 1945 0.6185
Years -1.6772** 0.6996 -1 .2011 0.7939
Eprears 0.3241 0.4701 0.3647 0.5353
Income N/A N/A 0.7795 0.5237
OIncome N/A N/A -0.8439 0.6090
Labor N/A N/A 0.9138 0.9739
OFeed N/A N/A -2.3037*** 0.8929
Cows -0.5654 0.6076 -1 2442* 0.7390
Auctions 0.1342 1.8112 -0.8945 2.1168
PTagS N/A N/A 2.6699 3.2275
ENotches N/A N/A -2.2580 2.0966
Brand N/A N/A 0.2855 2.2782
Tattoo N/A N/A -0.9769 2.1039
MTagS N/A N/A 2.3780 1.9142
RFID -4.5168 3.1866 -6.2983* 3.6132
NoID N/A N/A 3.3593 6.0893
RegNAIS 9.8683 * ** 1.8799 -1.0590 2.1269
ManNAIS 0.8588 1.0591 3.3516*** 1.1843
CRetailers 0.0269 0.0389 N/A N/A
CProcessors -0.0834* 0.0491 N/A N/A
CFeedlots 0.0877” 0.0439 N/A N/A
Sigma 17.6699*** 0.6130 19.7337*** 0.6941
Rho 0.5674*** 0.0335

 

 

' Source: Calculated from Survey Data Estimation
° *** 1% Significance level, ** 5% significance level, * 10% significance level

122

Table 7.2a. NAIS Premise Registration Estimates

 

 

Parameter Estimate Standard Error
Intercept -0.7491 0.5984
Age -0.0006 0.0063
CB -0.3904 0.3416
NP -0.5257 0.3480
NW -0.9237** 0.3678
SE 0296] 0.3525
SP -0.3297 0.3688
Member 0.1707*** 0.0532
Education 0.0539 0.0428
Years -0.0233 0.0497
Eprears 0.0160 0.0323
Income 0.0262 0.0375
OIncome 0.0280 0.0448
Labor 0.0654 0.0699
OFeed 0.0524 0.0631
Cows -0.0153 0.0468
Auctions -0.2092 0.1301
PTags 0.3305 0.2286
ENotches -0.0123 0.1499
Brand 0.0183 0.1589
Tattoo -0.0572 0.1605
MTags -0.0473 0.1398
RFID l.3439*** 0.2669
NoID -0.1599 0.4660
ManNAIS 0.1997* * * 0.0694

 

- Source: Calculated from Survey Data Estimation
- *** 1% significance level, ** 5% significance level, * 10% significance level

123

Table 7.3a. Producers’ Beliefs Concerning Mandatory NAIS Estimates

 

Outcome=Yes

Outcome=Undecided

 

 

Parameter Coefficient Standard Error Coefficient Standard Error
Age 0.0162 0.0119 -0.0238** 0.0121
CB 1.1410* 0.6424 1.1594* 0.6437
NP 2.0026*** 0.6413 2.0076*** 0.6408
NW 2.5673*** 0.7119 3.1515*** 0.7088
SE 1.7576*** 0.6606 1.6500** 0.6649
SP 1.8062*** 0.6541 1.2624* 0.6680
Member -0.3157*** 0.1220 -0.6289*** 0.1230
Years -0.091 8 0.0876 0.1619* 0.0903
Eprears -0.0491 0.0595 -0.0338 0.0598
Income -0.0950 0.0726 -0.1562** 0.0708
Cows -0.1095 0.0796 -0.0967 0.0800
Auctions 0.0770 0.2593 0.9363* ** 0.2629
BRetailers -0.0082** 0.0039 -0.0048 0.0037
BProcessors -0.0045 0.0049 -0.0051 0.0046
BFeedlots 0.0068 0.0061 0.0013 0.0059
CRetailers 0.0021 0.0060 -0.0019 0.0070
CProcessors -0.0048 0.0077 -0.0143 0.0088
CFeedlots 0.0123* 0.0066 0.0201 *** 0.0068
InvMills -3.9950*** 0.5200 -4.7925*** 0.5140
Constant 2.8262*** 0.9958 4.1158*** 1.0059

 

0 Source: Calculated from Survey Data Estimation
0 Outcome = No is the base outcome
0 *** 1% significance level, ** 5% Significance level, * 10% significance level

124

Table 7.4a. Monitoring/Managing Disease Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.7272*** 0.4772
Age -0.0112** 0.0052
CB -0.1375 0.2724
NP 0.1556 0.2697
NW 0.1826 0.2780
SE 0.3863 0.2907
SP 0.0308 0.2855
Member 0.0802* 0.0449
Education 0.0632* 0.0361
Years -0.0022 0.0423
Eprears 0.0050 0.0268
Income 0.0002 0.03 14
OIncome 0.0271 0.0374
Labor -0.0422 0.0594
OFeed -0.0053 0.0531
Cows 0.0207 0.0377
Auctions -0.0913 0.1 105
RegNAIS 0.3085*** 0.1056
Limit 2 0.2999*** 0.0629
Limit 3 0.8274*** 0.0855
Limit 4 1.7192*** 0.0983

 

- Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

125

Table 7.5a. Maintaining Current Foreign Markets Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.2923 * * * 0.4517
Age -0.0105** 0.0052
CB 0.0250 0.2611
NP 0.3451 0.2579
NW 0.2158 0.2676
SE 0.3494 0.2755
SP 0.2756 0.2753
Member 0.0445 0.0434
Education 0.0422 0.0352
Years 0.0613 0.0403
Eprears 0.0129 0.0267
Income -0.0016 0.0309
OIncome 0.0506 0.0364
Labor 0.0009 0.0580
OFeed 0.0036 0.0512
Cows -0.0004 0.0373
Auctions -0.2832*** 0.1084
RegNAIS 0.1942* 0.1027
Limit 2 0.3099*** 0.0606
Limit 3 1.0052*** 0.0851
Limit 4 1.8927*** 0.0964

 

0 Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

126

Table 7.6a. Accessing Foreign Markets Estimates

 

 

Parameter Estimate Standard Error
Intercept l . 1488* * 0.4631
Age -0.0079 0.0052
CB -0.4193 0.2730
NP -0.2871 0.2682
NW -0.2834 0.2787
SE -0.0550 0.2861
SP -0.1939 0.2858
Member 0.0479 0.0431
Education 0.0821* * 0.0348
Years 0.0801** 0.0402
Eprears 0.0272 0.0266
Income -0.0001 0.0309
OIncome 0.0456 0.0362
Labor 0.0072 0.0580
OFeed 0.0225 0.0512
Cows 0.0325 0.0372
Auctions -0.3022*** 0.1079
RegNAIS 0.1532 0.1017
Limit 2 0.3288*** 0.0616
Limit 3 1.0888*** 0.0862
Limit 4 1.9926*** 0.0975

 

- Source: Calculated from Survey Data Estimation
° *" 1% significance level, ** 5% significance level, * 10% significance level

127

Table 7.7a. Increasing Consumer Confidence Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.5097*** 0.4570
Age -0.0123** 0.0052
CB 0.2151 0.2616
NP 0.3998 0.2575
NW 0.2601 0.2675
SE 0.6477** 0.2800
SP 0.3597 0.2765
Member 0.0654 0.0433
Education 0.0239 0.0358
Years 0.0260 0.0411
Eprears 0.0031 0.0272
Income 0.0274 0.0314
OIncome 0.0278 0.03 72
Labor -0.0222 0.0593
OFeed -0.0354 0.0525
Cows 0.0550 0.0375
Auctions -0.2988*** 0.1 109
RegNAIS 0.2089** 0.1040
Limit 2 0.3783*** 0.0681
Limit 3 0.9007*** 0.0867
Limit 4 1.8124*** 0.0989

 

° Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

128

Table 7.8a. Concerns Regarding System Cost Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.3577*** 0.4666
Age -0.0177*** 0.0052
CB -0.2443 0.2710
NP -0. 1284 0.2675
NW -0.0270 0.2772
SE 0.2777 0.2865
SP 0.1047 0.2846
Member -0.0300 0.0436
Education —0.0430 0.0357
Years 0.1195*** 0.0410
Eprears -0.0147 0.0265
Income 0.0157 0.0314
OIncome 0.0225 0.0366
Labor -0.0668 0.0585
OFeed 0.1435*** 0.0515
Cows 0.1380*** 0.0373
Auctions 0.1630 0.1084
RegNAIS -0.1579 0.103 8
Limit 2 0.3637*** 0.0803
Limit 3 0.9879*** 0.1034
Limit 4 l.9788*** 0.1143

 

° Source: Calculated from Survey Data Estimation
- *** 1% significance level, ** 5% significance level, * 10% significance level

129

Table 7.9a. Concerns Regarding System Liability Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.5339“ * 0.4600
Age -0.0238*** 0.0052
CB 0.2575 0.2588
NP 0.4208* 0.2546
NW 0.4099 0.2643
SE 0.6905" 0.2735
SP 0.5006* 0.2726
Member 0.0463 0.0438
Education 00084 0.0357
Years 0.1328*** 0.0407
Eprears -0.0294 0.0266
Income 0.0348 0.0314
OIncome -0.0320 0.0366
Labor -0.0192 0.0589
OFeed 0.0953* 0.0520
Cows 0.0931** 0.0376
Auctions -0.0244 0.1084
RegNAIS -0.0962 0.103 5
Limit 2 0.4819*** 0.0965
Limit 3 1.2359*** 0.1169
Limit 4 2.0984*** 0.1243

 

- Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

130

Table 7.10a. Concerns Regarding Confidentiality Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.0775 * * 0.4523
Age -0.0171*** 0.0051
CB 0.0935 0.2586
NP 0.2412 0.2554
NW 0.3143 0.2661
SE 0.3161 0.2722
SP 0.2215 0.2729
Member 0.0497 0.0428
Education 0.0137 0.0352
Years 0.1788*** 0.0401
Eprears 0.0083 0.0264
Income -0.0013 0.0309
OIncome -0.0670* 0.0360
Labor -0.1063* 0.0575
OFeed 0.0421 0.0514
Cows 0.0449 0.0366
Auctions 0.1217 0.1063
RegNAIS -0.0998 0.1019
Limit 2 0.3971 *** 0.0667
Limit 3 l.0472*** 0.0868
Limit 4 1.8063*** 0.0962

 

0 Source: Calculated from Survey Data Estimation
- *** 1% significance level, ** 5% significance level, * 10% significance level

131

Table 7.11a. Concerns Regarding System Reliability Estimates

 

 

Parameter Estimate Standard Error
Intercept 0.7583* 0.4572
Age -0.0121** 0.0050
CB 0.2580 0.2649
NP 0.5420“ 0.2622
NW 0.5815** 0.2713
SE 0.6705** 0.2785
SP 0.6074** 0.2794
Member 0.0478 0.0428
Education 00409 0.0349
Years 0.1203*** 0.0400
Eprears -0.0205 0.0261
Income 0.0703** 0.0309
OIncome -0.031 1 0.0359
Labor -0.0192 0.0573
OFeed 0.1013** 0.0512
Cows 0.0722” 0.0364
Auctions -0.0657 0.1059
RegNAIS -0.0204 0.1015
Limit 2 0.4431*** 0.0884
Limit 3 1.2418*** 0.1104
Limit 4 2.2128*** 0.1190

 

0 Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

132

Table 7.12a. Concerns Regarding Free-Riding Estimates

 

 

Parameter Estimate Standard Error
Intercept 0.9534** 0.4402
Age -0.0117** 0.0049
CB 0.0303 0.2571
NP 0.3629 0.2530
NW 0.0907 0.2620
SE 0.4036 0.2707
SP 0.1657 0.2691
Member 00123 0.0419
Education -0.0139 0.0342
Years 0.1043*** 0.0395
EprearS -0.0206 0.0255
Income 0.0044 0.0301
OIncome 0.0021 0.0354
Labor 0.0032 . 0.0567
OFeed 0.0905** 0.0496
Cows 0.0631* * 0.03 52
Auctions 0.0343 0.1036
RegNAIS 0.0450 0.0991
Limit 2 0.4650*** 0.0704
Limit 3 1.3574*** 0.0907
Limit 4 2.0850*** 0.0986

 

0 Source: Calculated from Survey Data Estimation

° *** 1% significance level, ** 5% significance level, * 10% significance level

133

Table 7.13a. Concerns Regarding System Failures Estimates

 

 

Parameter Estimate Standard Error
Intercept 0.4841 0.4466
Age -0.0101** 0.0050
CB 0.2382 0.2557
NP 0.4932* 0.2520
NW 0.4913* 0.2624
SE 0.5469” 0.2698
SP 0.6192** 0.2706
Member 0.0238 0.0424
Education 0.0373 0.0350
Years 0.1034*** 0.0397
Eprears -0.0101 0.0262
Income 0.0268 0.0305
OIncome 0.0134 0.0360
Labor 0.0283 0.0580
OFeed 0.1242* * 0.0505
Cows 0.0362 0.0362
Auctions 0.0788 0.1051
RegNAIS -0.1479 0.1010
Limit 2 0.4027*** 0.0788
Limit 3 l.1646*** 0.1009
Limit4 2.1296*** 0.1101

 

0 Source: Calculated from Survey Data Estimation

- *** 1% significance level, ** 5% significance level, * 10% significance level

134

Table 7.14a. Implementation of COOL Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.6611*** 0.4426
Age -0.0044 0.0049
CB -0.2619 0.2611
NP -0.2461 0.2572
NW -0.0887 0.2665
SE 0.0258 0.2737
SP -0.1818 0.2719
Member -0.0656 0.0408
Education —0.0749** 0.0340
Years 0.0959“ 0.0387
Eprears 0.0006 0.0253
Income -0.0160 0.0296
OIncome -0.0883** 0.0348
Labor 0.0641 0.0559
OFeed 0.0101 0.0497
Cows 0.0027 0.0349
Auctions 0.2366" 0.1027
RegNAIS -0.2973*** 0.0985
Limit 2 0.7123*** 0.0695
Limit 3 1.4754*** 0.0832
Limit 4 2.1433*** 0.0926

 

0 Source: Calculated from Survey Data Estimation

° *** 1% significance level, ** 5% significance level, * 10% significance level

135

CHAPTER 8: CONCLUSIONS

With traceability becoming ever more important within the beef industry for
verification of animal health as well as marketing purposes, the need for traceability
systems that are attractive to producers as well as meet the goals that they were designed
for is evident. Traceability systems that are most aligned with the preferences of cow-
calf producers have the potential to experience higher voluntary participation.

Subsequently, a majority of the past research has sought to analyze consumer’s
perceptions towards voluntary traceability systems; however, there has been little
research on producer’s perceptions towards these same traceability systems. The
research presented within this thesis has illustrated the need for taking producers’
preferences into account when designing and imposing traceability systems.
Examinations of a mandatory versus voluntary National Animal Identification System
(N AIS) preference, self-revelation of current NAIS participation, and the most current
concerns and important issues to cow-calf producers regarding traceability provided
information that was useful in determining strategies that may enhance participation rates
and targeted marketing efforts of existing or potential live animal traceability systems.

Results indicate that producers from the Northwest (AZ, CA, CO, ID, MT, NV,
NM, OR, UT, WA, WY), relative to the Northern Crescent (MN, WI, CT, ME, MD MA,
NH, NJ, NY, PA, RI, VT), are less likely to have their premises registered with NAIS.
As such, this identifies an area that the United States Department of Agriculture (USDA)
may target to help increase voluntary premises registration in NAIS. Results concerning
mandatory versus voluntary NAIS preferences revealed that producers who use local

auctions to market their cattle are more likely undecided concerning mandatory NAIS.

136

This conclusion reveals that support for mandatory NAIS could be achieved by
promoting within or working with auction markets. Furthermore, cow-calf operators
with more years of experience are generally more concerned with issues related to animal
traceability implementation.

This thesis also utilized models that examined cow-calf producers’ preferences for
traceability systems and attributes which were characterized by heterogeneity. Thereby,
allowing us to better identify the success of alternative voluntary traceability systems that
could potentially exist in the beef industry. This “potential success” was measured by
producers’ preferences for certain traceability systems and system attributes.

As was implied by the latent classification (LCM) and random parameter Logit
(RPL) models, heterogeneity does exist between cow-calf producers and their preferences
for traceability systems and system attributes. Producers were found to be sensitive to
both price and information requirements in comparing voluntary traceability systems.
This implies that voluntary participation in traceability programs deemed undesirable to
the representative cow-calf producer may be accomplished by offering premiums for
participation; it also suggests participation may be enhanced by markets penalizing non-
participants. Furthermore, removal of traceability options contributes negatively to the
economic welfare of producers. Especially, for those producers who prefer No
Traceability. Welfare effects of imposing mandatory NAIS vary widely across producers
and model specifications. The LCM model suggests a large range for these welfare
effects of imposing mandatory traceability, where welfare estimates from the MNL and
RPL models fell in between this range. This provides strong evidence for the priority of

correct model specification.

137

Given cow-calf producers are sensitive to price adjustments and this thesis only
considered price adjustments as a single factor, further research could determine if
marginal utilities of discounts (penalties for nonparticipation) were the same as marginal
utilities of premiums (incentives to participate), answering the question is “price/head”
symmetrical across premiums and discounts. Another issue that could be addressed
would investigate if and how preferences concerning traceability systems change if
producer’s choice set changes. This would involve mapping traceability system choice
probabilities under different scenarios. These scenarios may include mandatory NAIS or
voluntary market scenarios in which producers may chose discounts for non-
participation.

Because model specification was found to be very important for the significance
of point estimates, WTA estimates, and welfare effects, an out-Of-sample exercise
determining which model performed the highest across the sample would be beneficial.

Thereby, allowing for more validation to be given to the various estimates.

138

APPENDICES

139

APPENDIX 1: COW-CALF PRODUCER SURVEY

Confidential Survey — For Research Purposes Only

Survey — Production Practices and Potential Beef Traceability Systems. Conducted by
Michigan State University in Collaboration with BEEF Magazine

1. 1am: Male Female
2. Iam years old.

3. My state of residence is:

 

4. Please check any farm organizations in which you are a member:

 

13 State or local cattlemen’s group Cl National Farmers Union
[:1 Breed association Cl National Livestock Association
El National Farm Bureau Cl National Farmers Organization
El National Cattlemen’s Beef El Other (please list)

Association (N CBA)
Cl Ranchers and Cattlemen’s

Action Legal Fund (R-Calf)

5. The best description of my educational background is:
[:1 Did not attend college

C] Technical training (Certification or Associates Degree)
El Attended College, No Bachelor’s (BS. or BA.) Degree

Cl Bachelor’s (BS. or B.A.) College Degree

Cl Graduate or Professional Degree (M.S., Ph.D., D.V.M., Law School)

[:1 Other (please describe)

 

6. How many years have you been raising beef cattle?

21-25 years

26-30 years

Over 30 years

I don’t raise beef cattle

21-25 years
26-30 years
Over 30 years

$75,000-$99,999
$ 100,000-$ 124,999

[:1 Less than 5 years El
[:1 6-10 years Cl
CI 11-15 years El
[:1 16-20 years El

7. How many more years do you expect your operation to be raising beef cattle?
Cl Less than 5 years E]
El 6-10 years Cl
El 11-15 years El
CI 16-20 years

8. Please estimate your annual pre-tax household income:
[:1 Less than $25,000 [:1
Cl $25,000-$49,999 El
[3 $50,000-$74,999 D

140

$125,000 or more

10.

ll.

12.

l3.

14.

15.

16.

Approximately what portion of your household income is from off-farm sources?
El Less than 20% Cl 60%-79%
El 20%—39% El 80% or more
El 40%-59%

What portion of your operation’s labor is supplied by non-family, paid employees?
El Less than 25% El 26%-50% Cl 51%-75% [:1 Over 75%

What portion of your operation’s feed/forage needs is produced by your own farm?
Cl Less than 25% Cl 26%-50% Cl 51%-75% [:1 Over 75%

Do you market cattle/beef based upon the following claims (Check Yes or No)?
Implant free Yes El No Cl 100% grass-fed diets Yes [:1 No El
Antibiotic free Yes Cl No El Humanely raised Yes El No [:1
Natural Yes [:1 NO El .

Organic Yes C] No 13 Other (please list)

 

Which one of the following methods do you most frequently use in marketing your
operations output (please check only one)?

Direct to traders CI Local butcher
El Direct to consumers Cl Packing Plant
El Local auction El Special auctions
El Direct to backgrounding systems (clubs, associations, etc.)
[:1 Direct to feedlots Cl Video/Internet auctions

How many beef cows that calved were on hand in your operation on January 1, 2007:

El None [:1 200-299
Cl Less than 50 Cl 300-499
[:1 50-99 Cl 500-999
Cl 100-199 I] 1,000 or more

How many head of cattle did your operation sell at the following production stages in
2007?

Calves (# head) Yearlings (# head) Finished Cattle (# head)
If you market finished cattle do you have your own Yes I] No [I feedlot?

Which of the following animal identification methods do you currently use (check all
that apply)?

13 Plastic ear tag El Brucellosis or any other metal tag
El Ear notches Cl Electronic ear tag (RFID)

El Brand Cl None

Cl Tattoo CI Other (please list)

 

141

17. In designing a national, individual animal traceability system how important are the
following issues to the US. beef industry (please circle your answers where 1 =
Entirely Unimportant, 2 = Unimportant, 3 = Neutral, 4 = Important, 5 = Very
Important):

 

 

 

Monitoring/managing disease: 1 2 3 4 5
Increasing consumer confidence: 1 2 3 4 5
Enhancing marketability: 1 2 3 4 5
Maintaining current foreign markets: 1 2 3 4 5
Accessing foreign markets: 1 2 3 4 5
Improving on-farm management: 1 2 3 4 5
Managing the supply chain: 1 2 3 4 5
Enhancing food safety: 1 2 3 4 5

18. In designing a national, individual animal traceability system how concerned are you
regarding the following issues in the US. beef industry (where 1 = Entirely
Unconcerned, 2 = Unconcerned, 3 = Neutral, 4 = Concerned, 5 = Very Concerned):

 

 

Cost to participating producer: 1 2 3 4 5
Confidentiality of information: 1 2 3 4 5
Reliability of technology: 1 2 3 4 5
Liability to participating producer: 1 2 3 4 5
Non-participating firms benefiting: 1 2 3 4 5
Failure of system to meet stated goals: 1 2 3 4 5

19. Please circle your level of agreement with each of the following statements (Where 1
= Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree):

Implementing individual animal traceability systems:

“is more cost effective for larger cow-calf operations.” 1 2 3 4 5

 

“results in more liability for cow-calf producers than
cattle owners at other stages of production.” 1 2 3 4 5

 

“is unnecessary if COOL (Country-of-Origin Labeling)
was implemented nationally.” l 2 3 4 5

 

“as a mandated system is exaggerated in need.” 1 2 3 4 5

 

142

Following are 4 scenarios, each containing 3 different options for you to select
from where 2 options are voluntary traceability systems you could participate in. These
two systems would require you to: a) provide basic contact information on your
farm/premises and b) individually identify your livestock (prior to any sales transactions
or movement to other premises) using approved radio frequency identification devices
(RFID). The 3rd option (N0 Traceability) is an alternative where you choose not to
participate in either of the 2 presented traceability systems. For your information in
interpreting the alternative traceability options please carefully read the following
descriptions:

NAIS Traceability: refers to the National Animal Identification System (N AIS).
o The NAIS program consists of three components:

0 Premise registration requires you to provide basic contact information.

0 Animal identification provides you with a uniform numbering system for
identifying your animals and linking them to a birthplace/premises of
origin.

0 Animal tracing allows you to choose an animal tracking database (owned
and operated by private industry groups or States) and report certain
animal movements that might pose a significant risk of disease
transmission.

0 USDA ensures producers that it will protect private information from disclosure.

Advanced Traceability: Same basic participation requirements as NAIS Traceability but
also requires additional information which may include:
0 Age Verification
0 Production Practice Information (e. g., growth hormones, grass-fed diets,
etc.)
0 Performance/Genetic Information
0 Health Certifications/Vaccinations Records
0 Random verification audits would be required to further validate consistency
between on-farm practices and information maintained within the traceability
system.

The entity managing each traceability system may take one of three forms:
0 Government: Entity such as the USDA.
0 Private - Industry: Entity specializes in traceability specifically for the beef
industry.
0 Private - Not Industry: Entity specializes in traceability for multiple livestock
specres.

In addition to differences in requirements and goals of these alternatives, these
options differ in terms of the premium or discount (per head sold) you would receive by
selecting that alternative. These price adjustments to the market price range from
discounts of up to $15 per head to premiums of up to $15 per head. Negative numbers
indicate discounts and positive numbers indicate premiums.

143

Research studies have found people to overstate their willingness to participate (or
accept discounts) in hypothetical situations. It is important that you make your selections
as if you were actually facing these choices. _P_1ace an “X” in the “I choose” box, below

the option that yo; wogld choose from each of the following 4 scenarios.

 

 

 

 

 

 

 

 

 

 

 

 

Scenario 1
Attribute NAIS Traceability Advanced Traceability No Traceability
Premium/Discount ($/head) $0.00 $0.00 -$7.50
Managing Entity Gove‘mment Private - Not Industry
Age Verification,
Additional Information Production Practices, and
HealthNaccinations
I choose 1:1 El E1
Scenario 2
Attribute NAIS Traceability Advanced Traceability No Traceability
Premium/Discount (S/head) $0.00 $15.00 $0.00
Managing Entity Government Private - Industry
Age Verification,
Additional Information Production Practices, and
HealthNaccinations
I choose El El E1
Scenario 3
Attribute NAIS Traceability Advanced Traceability No Traceability
Premium/Discount (S/head) $0.00 $7.50 -$7.50
Managing Entity Government Government
Age Verification,
Additional Information Production Practices, and
Performance/Genetics
I choose El El [:1
Scenario 4
Attribute NAIS Traceability Advanced Traceability No Traceability
Premium/Discount (S/head) $0.00 $15.00 -$15.00
Managing Entity Government Private - Not Industry
. . . Age Verification and
Addltlonal Informatlon HealthNaccinations
I choose El C1 13

 

144

 

 

 

 

20. If given the opportunity involving real money and binding decisions, how certain are
you that you would make the selections you indicated in the previous scenarios?
Please circle one number on this certainty scale:

VeryUncertain l- 2 - 3 - 4 - 5 - 6 - 7 - 8 - 9 - 10 VeryCertain
21. If individual animal traceability systems were put in place, how do you think the

resulting benefits and costs would be distributed through the beef industry’s supply
chain? Please allocate the percentages each of the following sectors captures:

Benefits: Costs:
Retailers Retailers
Processors Processors
Feedlot Producers F eedlot Producers
Cow-calf Producers Cow-calf Producers
Sum 100% Sum 100%

22. Are your operation’s premise(s) currently registered with USDA in the NAIS
(National Animal Identification System)?
[:1 Yes El No

23. Do you believe that NAIS (as previously outlined) should be a mandatory system

requiring all US. cattle producers to participate?
Cl Yes Cl No El Undecided

24. If the NAIS system remains a system of voluntary participation, by December 31,
2008 what do you predict will be (please circle your answers):

The percentage of US. cow-calf operations with premises registered in NAIS.

0% - 10% - 20% - 30% - 40% - 50% - 60% - 70% - 80% - 90% - 100%

The percentage of US. cow-calf operations with premises registered in NAIS
AND using NAIS approved RFID animal identification devices on the percentage
of cattle leaving their premises.

0% - 10% - 20% - 30% - 40% - 50% - 60% - 70% - 80% - 90% - 100%

Thank you for your time in completing this survey. Your input will strengthen our
research and help us obtain more accurate conclusions. If you wish to add any comments
that might be useful in our research please feel free to do so here:

 

 

 

 

145

APPENDIX 2: VARIABLES USED IN REGRESIONS AND DIAGNOSTICS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abbreviation Definition
Age Producer's Age
CB Corn Belt Region (IL, IN, IA, MO, OH)
NC Northern Crescent (MN, WI, CT, ME, MD MA, NH, NJ, NY, PA, RI,
VT)
NP Northern Plains (KS, NE, ND, SD)
NW Northwest (AZ, CA, CO, ID, MT, NV, NM, OR, UT, WA, WY)
SE Southeast (AL, FL, GA, SC, KY, NC, TN, VA, WV, AK, LA, MS)
SP Southern Plains Region (OK, TX)
Member Number of Memberships in indust£y Organizations
Education Educational Background
Years Years Raising Beef Cattle
Eprears Expected Future Years Raising Beef Cattle
Income Annual Pre-tax Household Income
OIncome Proportion of Off-farm Household Income
Labor Proportion of Operation's Labor Supplied by, Paid Employees
OF eed Feed/Forage Produced On-Farrn
Cows Marketing Through a Local Auction
Auctions Beef Cows that Calved in 2007
PTags Plastic Ear Tag Animal Identification
ENotches Ear Notches Animal Identification
Brand Brand Animal Identification
Tattoo Tattoo Animal ID Identification
MTag Brucellosis or AnLOther Metal Tag Animal Identification
RFID Electronic Ear Tag (RF ID) Identification
NoID None Animal Identification
RegNAIS Operation's Premise(s) Currently Registered in NAIS
ManNAIS Belief NAIS Should be Mandatory
BRetailers Benefits to Retailers over Cow-Calf producers
BProcessors Benefits to Processors over Cow-Calf producers
BFeedlots Benefits to Feedlot Producers over Cow-Calf producers
CRetailers Costs to Retailers over Cow-Calf producers
CProcessors Costs to Processors over Cow-Calf producers
CF eedlots Costs to F eedlot Producers over Cow-Calf producers
IndvID Individual Animal Identification (PTags, RF ID)
GrplD Group Animal Identification (ENotches, Brand, Tattoo, MTags)
NotID No/Other Animal Identification
PD Premium or Discount per head sold
No Trace No Traceability
Adv Trace Advanced Traceability
Pvtl Private - Industry Managing Entity
PvtNI Private — Not Industry Managing Entity

 

146

 

VARIABLES USED IN REGRESIONS AND DIAGNOSTICS (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abbreviation Definition
PP Production Practice Information
PG Performance/Genetic Information
HR Health Certification/Vaccination Records
AVPP Age Verification & Production Practice Information
AVPG Age Verification & Performance/Genetic Information
AVHR Age Verification & Health Certification/Vaccination Records
PPPG Production Practice & Performance/Genetic Information
PPHR Production Practice Information & Health Records
PGHR Performance/Genetic Information & Health Records
AVPPPG Age Verification & Production Practice & Performance/Genetic
Information
AVPPHR Age Verification & Production Practice Information & Health
Certification/Vaccination Records
PPPGHR Production Practice & Performance/Genetic Information & Health
Certification/Vaccination Records
AVPPPGHR Age Verification & Production Practice & Performance/Genetic
Information & Health Records
AT_Cows Interaction of Advanced Traceabilit)I and number of beef cows
NT_Cows Interaction of No Traceability and number of beef cows
AT_Member Interaction of Advanced Traceability and number of industry
memberships
NTEMember Interaction of No Traceability and number of industry memberships

 

AT_Auctions

Interaction of Advanced TraceabiliLy and local auction use

 

NLAuctions

Interaction of No Traceability and local auction use

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

AT_YearS Interaction of Advanced Traceability and years raising beef cattle
NT_YearS Interaction of No Traceability and years raising beef cattle
AT_IndID Interaction of Advanced TraceabiligI and individual ID use
NT_IndID Interaction of No Traceability and individual ID use
ALGrpID Interaction of Advanced Traceability and group ID use
NT_GrpID Interaction of No Traceability and group ID use
AT_NOID Interaction of Advanced Traceability and no/other ID use
NT_NOID Interaction of No Traceability and no/other ID use
AT_CB Interaction of Advanced Traceability and Corn Belt
NT_yCB Interaction of No Traceability and Corn Belt
AT_NP Interaction of Advanced Traceability and Northern Plains
NT_NP Interaction of No Traceability and Northern Plains
AT_NW Interaction of Advanced Traceability and Northwest
N'LNW Interaction of No Traceability and Northwest
AT_SE Interaction of Advanced Traceability and Southeast
NT_[SE Interaction of No Traceability and Southeast
AT_SP Interaction of Advanced Traceability and Southern Plains
NT_SP Interaction of No Traceability and Southern Plains

 

147

 

APPENDIX 3: COW-CALF PRODUCER SURVEY DEMOGRAPHIC,
PERCEPTION, AND CHOICE EXPERIMENT SUMMARY TABLES

3.1 Demographics

Table A.3.]. Producers’ Gender (Survey Question 1)

 

 

Number Reporting Percent RemrtinL
Male 602 92.47%
Female 49 7.53%
Total 65 1 100.00%

 

Source: Calculated from Data

Table A.3.2. Producers’ Age (Survey Question 2)

 

 

N N Standard
Valid Missing Mean Median Deviation Minimum Maximum
646 9 57.51 58 13.46 17 90

 

Source: Calculated from Data

Table A.3.3. Producers’ State of Residence (Survey Question 3)

 

 

Number Percent Number Percent
Reporting RQorting Reporting Reporting_

Alabama 19 2.93% Nebraska 5 l 7.86%
Arizona 3 0.46% Nevada 7 1.08%
Arkansas 6 0.92% New Mexico 5 0.77%
California 19 2.93% New York 3 0.46%
Colorado 20 3.08% North Carolina 7 1.08%
Florida 3 0.46% North Dakota 34 5.24%
Georgia 9 1.39% Ohio 8 1.23%
Hawaii I 0.15% Oklahoma 36 5.55%
Idaho 9 I .39% Oregon 10 1 54%
Illinois 12 1.85% South Carolina 2 0.31%
Indiana 3 0.46% South Dakota 50 7.70%
Iowa 44 6.78% Tennessee 14 2. 16%
Kansas 45 6.93% Texas 62 9.55%
Kentucky 10 1.54% Utah 10 1.54%
Louisiana 3 0.46% Virginia 13 2.00%
Maryland I 0.15% Washington 6 0.92%
Minnesota 10 1.54% West Virginia 2 0.31%
Mississippi 4 0.62% Wisconsin 10 1.54%
Missouri 42 6.47% Wyoming 23 3.54%
Montana 33 5.08% Total 649 100.00%

 

Source: Calculated from Day

 

28 Michigan was removed from the data due to the mandatory nature of their beef traceability system.

148

Table A.3.4. States within the ERS Farm Production Regions

 

Appnlecme
Kentucky

North Carolina
Tennessee
Virginia
West Virginia

Mountain
Arizona

Colorado
Idaho
Montana
Nevada
New Mexico
Utah
Wyoming

Corn Belt
Illinois
Indiana

Iowa

Missouri

Ohio

MM
Connecticut
Delaware
Maine
Maryland
Massachusetts
New Hampshire
New Jersey
New York
Pennsylvania
Rhode Island
Vermont

Delta States Northern Plains Southern Plains

 

 

Arkansas Kansas Oklahoma
Louisiana Nebraska Texas
Mississippi North Dakota
South Dakota
Pacific Southeast Lake States
California Alabama Minnesota
Oregon Florida Wisconsin
Washington Georgia

South Carolina

 

Table A.3.5. ERS Farm Production Regions

 

Number Reporting

Percent Reporting

 

 

Appalachia 46 7.10%
Corn Belt 109 16.82%
Delta States 13 2.01%
Lake States 20 3.09%
Mountain 1 10 16.98%
Northeast 4 0.62%
Northern Plains 180 27.78%
Pacific 35 5.40%
Southeast 34 5.25%
Southern Plains 97 14.97%
Total 21(5548 100.00%

Source: Calculated from Data

 

2'9 Base is equal to N-I because Hawaii was not assigned a production region.

149

Table A.3.6. States within the U.S. Regions

 

 

Northwest Northern Crescent
Arizona Minnesota
Colorado Wisconsin
Idaho Connecticut
Montana Delaware
Nevada Maine
New Mexico Maryland
Utah Massachusetts
Wyoming New Hampshire
California New Jersey
Oregon New York
Washington Pennsylvania
Rhode Island
Vermont

Southeast
Alabama
Florida
Georgia
South Carolina
Kentucky
North Carolina
Tennessee
Virginia
West Virginia
Arkansas
Louisiana
Mississippi

MM
Kansas
Nebraska
North Dakota
South Dakota

Southern Plains
Oklahoma
Texas

Corn Belt
Illinois
Indiana

Iowa

Missouri

Ohio

U.S. Regions reflect ERS Farm Resource Regions and ERS Production Regions30

Table A.3.7. U.S. Regions

 

Southeast Area (Appalachia, Southeast, Delta States)
Northern Crescent (Lake States, Northeast)

Northwest Area (Pacific, Mountain)

Corn Belt
Northern Plains
Southern Plains

Number Reporting Percent Reporting

93 14.35%

24 3 .70%

145 22.38%

109 16.82%

180 27.78%

97 14.97%

648 100.00%

Total

 

Source: Calculated from Data3r

Table A.3.8. Farm Organizations of Membership (Survey Question 4)

 

 

Number Reporting Percent Reporting_
XE M T0_wl XE M
State or local cattlemen's group 338 317 655 51.60% 48.40%
Breed association 131 524 655 20.00% 80.00%
National Farm Bureau 225 430 655 34.35% 65.65%
National Cattlemen's Beef Association 145 510 655 22.14% 77.86%
Rancher's and Cattlemen’s Action Legal Fund 79 576 655 12.06% 87.94%
National Farmers Union 41 614 655 6.26% 93.74%
National Livestock Association 17 638 655 2.60% 97.40%
National Farmers Organization 12 643 655 1.83% 98.17%
Other 5 l 604 655 7.79% 92.21%
Total 1039 4856 5895

 

Source: Calculated from Data; Percents may reflect multiple answers

 

’0 The initial set was condensed for ease in estimation and interpretation.

3' Base is equal to N-l because Hawaii was not assigned a production region.

150

Table A.3.8. Farm Organizations of Membership (cont’d)
Others listed:

Alabama Farmer's Federation
American Cattlemen's Association
American Soybean Association
National Corn Growers Association
American Quarter Horse Association
American Sheep Industry

Beef Quality Assurance

Cattle Women

Colorado Holistic Management
Conservation District

County Natural Beef

Dekalb County Cattlemen's Assoc.
Family Farms for the Future

FF A Alumni

Georgia Farm Bureau

I-Cow

Independent Cattlemen's Association.
Local Fair Board

Local Las Animas Company Livestock

Mercy for Animals

Montana Association of Conservation Districts
Natural Resources Conservation Service

Montana Organic Organization

Big Horn Company Livestock Association
Nebraska Cattlemen

Concordia Institute

Northern Plains Sustainable Agriculture Society
Organic Grassfed Beef Association

Organic Farmers Agency for Relationship Marketing
Weston A Price Foundation

State Farm Bureau

Texas Southwest Cattle Raisers

Texas Cattle Feeders Association

Texas and Southwestern Cattle Raisers Association
United States Cattlemen’s Association

Table A.3.9. Educational Description (Survey Question 5)

 

 

Number Percent
Reporting Reporting
Did not attend college 204 31.34%
Technical training (Certification or Associates Degree) 49 7.53%
Attended College, No Bachelor's (BS. or BA.) Degree 1 18 18.13%
Bachelor's (BS. or B.A.) College Degree 194 29.80%
Graduate or Professional Degree (M.S., Ph.D., D.V.M., Law School) 80 12.29%
Other 6 0.92%
Total 65 1 100.00%

 

Source: Calculated from Data

Others listed:

0 U.S. Army Training
0 Farm Management (8 week course)

0 Military
0 U.S. Navy Training

151

Table A.3.10. Years Raising Cattle (Survey Question 6)
Number Reporting Percent Reporting_

 

 

Less than 5 years 7 1.08%
6-10 years 23 353%
11-15 years 29 445%
16-20 years 34 5.22%
21-25 years 56 8.60%
26-30 years 65 9.98%
Over 30 years 427 65.59%
I don’t raise beef cattle 10 1.54%
Total 65 1 100.00%

 

Source: Calculated from Data

Table A.3.11. Expected Future Years Raising Cattle (Survey Question 7)

 

Number Rgeporting Percent Reporting_

 

Less than 5 years 74 1 1, 56%
6—10 years 109 17.03%
“-15 years 75 11.72%
l6-20 years 95 14.84%
21-25 years 45 7.03%
26-30 years 23 3.59%
Over 30 years 219 34.22%
Total 640 100.00%

 

Source: Calculated from Data

Table A.3.12. Estimated Annual Pre-tax Household Income (Survey Question 8)

 

Number Reporting Percent Reporting_

 

Less than $25,000 30 4.96%
$25,000-$49,999 106 17.52%
$50,000-$74,999 134 22.15%
$75,000-$99,999 96 15.87%
$ 100,000-$ 124,999 62 10.25%
$125,000 or more 177 29.26%
Total 605 100.00%

 

Source: Calculated from Data

Table A.3.13. Off-Farm Household Income (Survey Question 9)

Number Reporting Percent Reporting_

 

 

Less than 20% 288 45.50%
20%-39% 95 15.01%
40%-59% I 19 18.80%
60%-79% 54 8.53%
80% or more 77 12.16%
Total 633 100.00%

 

Source: Calculated from Data

152

Table A.3.l4. Labor Supplied by Paid Employees (Survey Question 10)

 

 

Number Reportigg Percent Reporting_
Less than 25% 480 75.00%
26%—50% 73 l 1.41%
51%-75% 43 6.72%
Over 75% 44 6.88%
Total 640 100.00%

 

Source: Calculated fi'om Data

Table A.3.15. Feed Needs Produced by Own Farm (Survey Question 11)
Number ReportinL Percent Reporting_

 

Less than 25% 55 8.59%
26%-50% 57 8.91%
51%-75% 104 16.25%
Over 75% 424 66.25%
Total 640 100.00%

 

Source: Calculated from Data

Table A.3.16. Production Claims (Survey Question 12)

 

 

Number Repormg Percent Rgportinj_

$6.8 1L0 LL81 Yes N0.
Implant free 260 303 563 46.18% 53.82%
Antibiotic free 139 387 526 26.43% 73.57%
Natural I35 366 501 26.95% 73.05%
Organic 17 432 449 3.79% 96.21%
100% grass-fed diets 95 402 497 19.11% 80.89%
Humanely raised 283 237 520 54.42% 45.58%
Other 33 0 33 100.00% 0.00%

Total 962 2127 3089

 

Source: Calculated from Data; Percents may reflect multiple answers

Others listed:
0 Age & Source Verified
0 Angus Breeding Cattle
o Backgrounded
0 Beef Quality Assurance
0 Boostered
0 Calves right off summer grass
0 Calves sold off cows, no weaning
- Dewormed
0 Feeders direct, not from auction
0 F VO Certified Organic
0 Genetics
0 Grain fed
0 Johne's Disease Negative
0 Marketing Co-op
0 Never sick

0 Non-hormone

0 Pets

0 Preconditioned

o Predator friendly

0 Rangeland

0 Seed stock

0 Some creep feeding

a They are delicious (Local Slaughter)
o Vaccinated

0 Vaccinated 9 way

0 Vaccinated PR3

o Vaccinated against calf hood diseases
0 Weaned

0 Winter Hay & Supplements

153

Table A.3.17. Marketing Methods (Survey Question 13)
Number Reporting Percent Repprtirg

 

Direct to traders 77 9.73%
Direct to consumers 34 4.30%
Local auction 384 48.55%
Direct to backgrounding systems 23 2.91%
Direct to feedlots 70 8.85%
Local slaughter house 13 1.64%
Packing Plant 79 9.99%
Special auctions (clubs, associations, etc.) 39 4.93%
Video/lntemet auctions 72 9.10%
Total 791 100.00%

 

Source: Calculated from Data; Percents may reflect multiple answers

Table A.3.18. Beef Cows that Calved (Survey Question 14)

Number Reporting Percent Reporting_

 

 

None 27 4.24%
Less than 50 37 5.81%
50-99 1 15 18.05%
100-199 209 32.81%
200-299 1 16 18.21%
300-499 65 10.20%
500-999 48 7.54%
1,000 or more 20 3.14%
Total 637 100.00%

 

Source: Calculated from Data

Table A.3.19. Operation’s Sales at Production Stages (Survey Question 15)

 

 

N N Standard
Valid Missing Mean Median Deviation Minimum Maximum
Calves 457 198 172.53 100 336.60 0 5000
Yearlings 315 340 216.01 60 514.22 0 5000
Finished Cattle 237 418 440.95 60 1437.33 0 15000

 

Source: Calculated from Data

Table A.3.20. Market Finished Cattle and Own Feedlot32 (Survey Question 15)

 

 

Number Reporting Percent Reporting
E E9 _Total XE N9
129 283 412 31.31% 68.69%

 

Source: Calculated from Data

 

’2 These results are not actually accurate because many producers indicated that they didn’t own there own
feedlot even if they did not market finished cattle in 2007.

154

Table A.3.21. Current Animal Identification Methods (Survey Question 16)

 

 

Number Reporting Percent Reporting__
Leg E9 Mal __Y_8§ N9
Plastic ear tag 566 78 644 87.89% 12.11%
Ear notches 162 482 644 25.16% 74.84%
Brand 377 266 643 58.63% 41.37%
Tattoo 128 515 643 19.91% 80.09%
Brucellosis or any other metal tag 244 399 643 37.95% 62.05%
Electronic ear tag (RF ID) 59 584 643 9.18% 90.82%
None 18 625 643 2.80% 97.20%
Other 8 635 643 1.24% 98.76%

 

Source: Calculated from Data

Others listed:

0 Bangs

o Brisket tags

0 FVO Certification Records
0 Green tag

0 Registration Number

0 Sure Health Tags

- Wattle tag

Table A.3.22. Current Premise Registration in NAIS (Survey Question 22)

 

 

Number Reporting Percent Reporting
Yes. U2 _Total Yes M
275 350 625 44.00% 56.00%

 

Source: Calculated from Data

155

3.2 Perceptions

Table A.3.23. Issues in Designing a Traceability System (Survey Question 17)

 

 

N N Standard

Valid Missing Mean Median Deviation
Monitoring/managing disease 618 37 4.10 4 1.13
Increasing consumer confidence 608 47 4.03 4 1.16
Enhancing marketability 608 47 3.82 4 1.17
Maintaining current foreign markets 607 48 3.89 4 1.17
Accessing foreign markets 606 49 3.83 4 1.18
Improving on-farm management 605 50 3.49 4 1.29
Managing the supply chain 606 49 3.44 4 1.24
Enhancing food safety 617 38 3.95 4 1.21

 

Source: Calculated from Data

Table A.3.23a. Monitoring/Managing Disease (Survey Question 17)

 

Number Rgmrting Percent Reporting

 

Entirely Unimportant 36 5.83%
Unimportant 27 4.37%
Neutral 70 l 1.33%
Important 192 3 I .07%
Very Important 293 47.41%
Total 6 1 8 100.00%

 

Source: Calculated from Data

Table A.3.23b. Increasing Consumer Confidence (Survey Question 17)

 

Number ReportinL Percent Reporting_

 

Entirely Unimportant 36 5.92%
Unimportant 35 5.76%
Neutral 75 12.34%
Important 190 31 25%
Very Important 272 44.74%
Total 608 100.00%

 

Source: Calculated from Data

Table A.3.23e. Enhancing Marketability (Survey Question 17)
Number Reporting Percent ReportingL

 

 

Entirely Unimportant 44 7.24%
Unimportant 36 5.92%
Neutral 105 17.27%
Important 222 36.51%
Very Important 201 33.06%
Total 608 100.00%

 

Source: Calculated from Data

156

Table A.3.23d. Maintaining Current Foreign Markets (Survey Question 17)

 

Number Reporting Percent Reportiri

 

Entirely Unimportant 42 6.92%
Unimportant 32 5.27%
Neutral 107 17.63%
Important 196 32.29%
Very Important 230 37.89%
Total 607 100.00%

 

Source: Calculated from Data

Table A.3.23e. Accessing Foreign Markets (Survey Question 17)

 

Number Reporting Percent Reporting_

 

Entirely Unimportant 44 7.26%
Unimportant 32 5.28%
Neutral 120 19.80%
Important 197 32.51%
Very Important 213 35.15%
Total 606 100.00%

 

Source: Calculated from Data

Table A.3.23f. Improving On-Farm Management (Survey Question 17)

 

Number Reporting Percent Reporting_

 

Entirely Unimportant 71 1 1.74%
Unimportant 58 9.59%
Neutral 140 23.14%
Important 178 29.42%
Very Important 158 26. 12%
Total 605 100.00%

 

Source: Calculated from Data

Table A.3.23g. Managing the Supply Chain (Survey Question 17)

Number Reporting Percent ReportingL

 

 

Entirely Unimportant 66 10.89%
Unimportant 56 9.24%
Neutral 166 27.39%
Important 182 30.03%
Very Important 136 22.44%
Total 606 100.00%

 

Source: Calculated from Data

157

Table A.3.23b. Enhancing Food Safety (Survey Question 17)

Number Reporting Percent Reporting_

 

 

Entirely Unimportant 46 7.46%
Unimportant 32 5.19%
Neutral 96 15.56%
Important 174 28.20%
Very Important 269 43.60%
Total 6 1 7 100.00%

 

Source: Calculated from Data

Table A.3.24. Concerns in Designing a Traceability System (Survey Question 18)

 

 

N N Standard

Valid Missing Mean Median Deviation
Cost to participating producer 630 25 4.19 4 1.02
Confidentiality of information 623 32 3.97 4 1.20
Reliability of technology 625 30 4.05 4 1.03
Liability to participating producer 623 32 4.16 4 1.03
Non-participating firms benefiting 622 33 3.73 4 1.18
Failure of system to meet stated goals 619 36 3.99 4 1.06

 

Source: Calculated from Data

Table A.3.24a. Cost to Participating Producer

 

Number Reporting Percent Eporting

 

Entirely Unconcemed 24 3.81%
Unconcemed 20 3. 17%
Neutral 73 1 1.59%
Concerned 207 32.86%
Very Concerned 306 48.57%
Total 630 100.00%

 

Source: Calculated from Data

Table A.3.24b. Confidentiality of Information

 

Number Reportig Percent Reporting_

 

Entirely Unconcemed 42 6.74%
Unconcemed 35 5.62%
Neutral 101 16.21%
Concerned 164 26.32%
Very Concerned 281 45.10%
Total 623 100.00%

 

Source: Calculated from Data

158

Table A.3.24c. Reliability of Technology
Number Reporting Percent Reporting_

 

 

Entirely Unconcemed 22 3.52%
Unconcemed 24 3.84%
Neutral 1 10 17.60%
Concerned 213 34.08%
Very Concerned 256 40.96%
Total 625 100.00%

 

Source: Calculated from Data

Table A.3.24d. Liability to Participating Producer
Number Remrting Percent Reporting_

 

 

Entirely Unconcemed 20 3.21%
Unconcemed 25 4.01%
Neutral 92 14.77%
Concerned 1 83 29.37%
Very Concerned 303 48.64%
Total 623 100.00%

 

Source: Calculated from Data

Table A.3.24c. Non-Participating Firms Benefiting
Number Reporting Percent Reporting_

 

 

Entirely Unconcemed 41 6.59%
Unconcemed 43 6.91%
Neutral 162 26.05%
Concerned 171 27.49%
Very Concerned 205 32.96%
Total 622 1 00.00%

 

Source: Calculated from Data

Table A.3.24f. Failure of System to Meet Stated Goals
Number Reporting Percent Reporting_

 

 

Entirely Unconcemed 27 4.36%
Unconcemed 26 4.20%
Neutral 109 17.61%
Concerned 219 35.38%
Very Concerned 238 38.45%
Total 61 9 100.00%

 

Source: Calculated from Data

159

Table A.3.25. Common Traceability Statements (Survey Question 19)

 

N N Standard
Valid Missifi Mean Median Deviation
“is more cost effective for larger cow-calf

operations.” 619 36 3 .28 3 1.19
“results in more liability for cow-calf producers

than cattle owners at other stages of production.” 619 36 3.73 4 1.14
“is unnecessary if COOL (Country-of-Origin

Labeling) was implemented nationally.” 619 36 3.37 3 1.28
“as a mandated system is exaggerated in need.” 618 37 3.38 3 1.23

 

Source: Calculated from Data

Table A.3.25a. “Is More Cost Effective for Larger Cow-Calf Operations”
Number Reporting Percent Reporting

 

Strongly Disagree 61 9.85%
Disagree 87 14.05%
Neutral 192 31.02%
Agree 173 27.95%
Strongly Agree 106 17.12%
Total 619 100.00%

 

Source: Calculated from Data

Table A.3.25b. “Results in More Liability for Cow-Calf Producers”

Number Reporting Percent Reporting_

 

Strongly Disagree 27 4.36%
Disagree 69 1 1.15%
Neutral 136 21.97%
Agree 198 31.99%
Strongly Agree 189 30.53%
Total 619 100.00%

 

Source: Calculated from Data

Table A.3.25c. “Is Unnecessary if COOL was Implemented Nationally”

 

Number Reporting Percent Reporting_

 

Strongly Disagree 61 9.85%
Disagree 97 15.67%
Neutral 167 26.98%
Agree 14 1 22.78%
Strongly Agree 153 24.72%
Total 619 100.00%

 

Source: Calculated from Data

160

Table A.3.25d. “As a Mandated System is Exaggerated”
Number Reporting Percent Reporting_

 

 

Strongly Disagree 54 8.74%
Disagree 87 14.08%
Neutral 191 30.91%
Agree 143 23.14%
Strongly Agree 143 23.14%
Total 61 8 100.00%

 

Source: Calculated from Data

Table A.3.26. Allocation of Benefits with Traceability (Survey Question 21)

 

 

Benefits to: N Valid N Missing Mean Median Standard Deviation
Retailers 514 141 34.38 30 24.93
Processors 513 142 26.56 25 20.52
Feedlot Producers 512 143 18.92 20 15.67
Cow-calf Producers 515 140 19.80 10 22.28

 

Source: Calculated from Data

Table A.3.27. Allocation of Costs with Traceability (Survey Question 21)

 

 

Costs to: N Valid N MissinL Mean Median Standard Deviation
Retailers 51 1 144 1 1.22 5 17.21
Processors 5 1 1 144 12.40 10 14.10
Feedlot Producers 5 1 1 144 18.71 20 15.00
Cow-calf Producers 518 137 58.36 60 29.22

 

Source: Calculated from Data

Table A.3.28. Should NAIS be Mandatory (Survey Question 23)

 

 

Number Reporting Percent Reporting
jigs _Ng Undecided Total m _N_Q Undecided
132 313 177 622 21 .22% 50.32% 28.46%

 

Source: Calculated from Data

Table A.3.29. Forecast of Premises Registered (Survey Question 24)

 

N Valid N Missing Mean Median Standard Deviation
593 62 40.47 40 18.87
Source: Calculated from Data

 

Table A.3.30. Forecast of Premises Registered and RFID Use (Survey Question 24)

N Valid N Missing Mean Median Standard Deviation
577 78 31.58 30 20.11
Source: Calculated from Data

 

 

161

3.3 Choice Experiment

Table A.3.31. General Choice Experiment Statistics

 

Returned Not Returned complete Complete%

 

Block A 136 264 91 66.91%
Block B 123 277 97 78.86%
Block C 127 273 97 76.38%
Block D 135 265 109 80.74%
Block E 134 266 100 74.63%

Total 655 1345 494 75.42%

 

Source: Calculated from Data

Table A.3.32. Complete Choice Experiments

 

 

NAIS Adv No

NAIS Trace Trace % Adv Trace Trace % No Trace Trace % Total

Block A 92 25.27% 183 50.27% 89 24.45% 364
Block B 124 31.96% 141 36.34% 123 31.70% 388
Block C 90 23.20% 154 39.69% 144 37.1 1% 388
Block D 152 34.86% 138 31.65% 146 33.49% 436
Block E 128 32.00% 175 43.75% 97 24.25% 400
Total 586 29.66% 791 40.03% 599 30.31% 1976

 

Source: Calculated from Data

162

Table A.3.33. Traceability Scenarios Block A
Traceability Scenarios Block A

 

 

Number Percent
Scenario I ReponiLL Reporting_
NAIS Traceability ($0.00, Government) 26 25.24%
Advanced Traceability ($0.00; Private - Not Industry; Age Verification,
Production Practices, and Health/Vaccinations) 55 53.40%
No Traceability (-$7.50) 22 21.36%
Total 103 100.00%
Scenario 2
NAIS Traceability ($0.00, Government) 20 19.23%
Advanced Traceability ($15.00; Private — Industry; Age Verification,
Production Practices, and Health/Vaccinations) 55 52.88%
No Traceability ($0.00) 29 27.88%
Total 1 04 100.00%
Scenario 3
NAIS Traceability ($0.00, Government) 29 30.53%
Advanced Traceability ($7.50; Government; Age Verification, Production
Practices, and Performance/Genetics) 36 37.89%
No Traceability ($-7.50) 30 31.58%
Total 95 100.00%
Scenario 4
NAIS Traceability ($0.00, Government) 22 22.45%
Advanced Traceability ($15.00; Private - Not Industry; Age Verification
and Health/Vaccinations) 61 62.24%
No Traceability ($-7.50) 15 15.31%
Total 98 100.00%

 

Source: Calculated from Data

163

Table A.3.34. Traceability Scenarios Block B
Traceability Scenarios Block B

 

 

Number Percent
Scenario 1 Reporting Reportiri
NAIS Traceability ($0.00, Government) 36 33.96%
Advanced Traceability ($7.50; Private - Not Industry; Age Verification and
Performance/Genetics) 39 36.79%
No Traceability (-$7.50) 31 29.25%
Total 106 100.00%
Scenario 2
NAIS Traceability ($0.00, Government) 26 24.53%
Advanced Traceability ($0.00; Private - Industry; Age Verification and
HealthNaccinations) 47 44.34%
No Traceability ($0.00) 33 31.13%
Total 1 06 100.00%
Scenario 3
NAIS Traceability ($0.00, Government) 36 36.36%
Advanced Traceability ($0.00; Government; Production Practices and
Performance/Genetics) 29 29.29%
No Traceability (-$15.00) 34 34.34%
Total 99 100.00%
Scenario 4
NAIS Traceability ($0.00, Government) 29 28.43%
Advanced Traceability ($7.50; Private - Industry; Production Practices and
HealthNaccinations) 49 48.04%
No Traceability (-$15.00) 24 23.53%
Total 102 100.00%

 

Source: Calculated from Data

164

Table A.3.35. Traceability Scenarios Block C

 

Traceability Scenarios Block C

 

Number Percent
Scenario 1 Report—ing ReportinL
NAIS Traceability ($0.00, Government) 28 27.18%
Advanced Traceability ($0.00; Private - Not Industry; Age Verification) 32 31.07%
No Traceability ($0.00) 43 41.75%
Total 103 100.00%
Scenario 2
NAIS Traceability ($0.00, Government) 20 20.00%
Advanced Traceability ($7.50; Government; Age Verification and
Production Practices) 37 37.00%
No Traceability ($0.00) 43 43.00%
Total 1 00 100.00%
Scenario 3
NAIS Traceability ($0.00, Government) 21 21.21%
Advanced Traceability ($7.50; Government; Age Verification) 42 42.42%
No Traceability (-$7.50) 36 36.36%
Total 99 100.00%
Scenario 4
NAIS Traceability ($0.00, Government) 25 25.25%
Advanced Traceability ($7.50; Private - Not Industry; Health/Vaccinations) 48 48.48%
No Traceability (-$15.00) 26 26.26%
Total 99 1 00.00%

 

Source: Calculated from Data

165

Table A.3.36. Traceability Scenarios Block D

 

Traceability Scenarios Block D

Scenario 1

NAIS Traceability ($0.00, Government)
Advanced Traceability ($15.00; Government; Production Practices and
HealthNaccinations)

No Traceability ($0.00)
Total

Scenario 2

NAIS Traceability ($0.00, Government)

Advanced Traceability ($15.00; Government; Age Verification, Production
Practices, Performance/Genetics, and Health/Vaccinations)

No Traceability (-$7.50)
Total

Scenario 3
NAIS Traceability ($0.00, Government)
Advanced Traceability ($7.50; Private - Not Industry; Production Practices,
Performance/Genetics, and Health/Vaccinations)
No Traceability ($0.00)
Total

Scenario 4
NAIS Traceability ($0.00, Government)
Advanced Traceability ($15.00; Private - Industry; Age Verification and
Production Practices)
No Traceability (-$7.50)
Total

 

Number Percent
Reporting Reporting_

34 30.09%
34 30.09%
45 39.82%

1 13 100.00%
48 42.86%
28 25.00%
36 32.14%

1 12 100.00%
37 32.46%
35 30.70%
42 36.84%

1 14 100.00%
35 31.25%
48 42.86%
29 25.89%

1 12 100. 00%

 

Source: Calculated from Data

166

Table A.3.37. Traceability Scenarios Block E
Traceability Scenarios Block E

 

 

Number Percent
Scenario 1 Reporting Reporting_
NAIS Traceability ($0.00, Government) 41 39.42%
Advanced Traceability ($0.00; Government; Production Practices,
Performance/Genetics, and Health/Vaccinations) 35 33.65%
No Traceability (-$15.00) 28 26.92%
Total 1 04 100.00%
Scenario 2
NAIS Traceability ($0.00, Government) 36 34.95%
Advanced Traceability ($0.00; Government; Performance/Genetics and
Health/Vaccinations) 38 36.89%
No Traceability (-$15.00) 29 28.16%
Total 103 100.00%
Scenario 3
NAIS Traceability ($0.00, Government) 27 25.71%
Advanced Traceability ($0.00; Private - Industry; Production Practices) 54 51.43%
No Traceability (-$7.50) 24 22.86%
Total 105 100.00%
Scenario 4
NAIS Traceability ($0.00, Government) 27 26.47%
Advanced Traceability ($7.50; Private - Industry; Performance/Genetics) 59 57.84%
No Traceability (-$15.00) 16 15.69%
Total 102 100.00%

 

Source: Calculated from Data

167

APPENDIX 4: SUPPLEMENTAL REGRESSIONS AND DIAGNOSTICS

“In designing a national, individual animal traceability system how important managing

the supply chain to the U.S. beef industry? ”

Table A.4.1. Managing the Supply Chain Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.6491 *** 0.4573
Age -0.0055 0.0049
CB -0.1182 0.2716
NP 0.2950 0.2680
NW 0.0002 0.2764
SE 0.4013 0.2852
SP 0.1373 0.2836
Member 0.0176 0.0418
Education -0.0443 0.0340
Years 0.0286 0.0392
Eprears 0.0025 0.0255
Income -0.0218 0.0298
OIncome 0.0075 0.0352
Labor -0.0091 0.0558
OFeed -0.1227** 0.0506
Cows 0.0659* 0.0355
Auctions -0.0402 0. 103 8
RegNAIS 0.0960 0.0988
Limit 2 0.4603*** 0.0606
Limit 3 1.2744*** 0.0796
Limit 4 2.1598*** 0.0925

0 *** 1% significance level, ** 5% significance level, * 10% significance level

Table A.4.2. ME to Managing the Supply Chain

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
OFeed 0.0199 0.0125 0.0148 -0.0120 -0.0352
Cows -0.0107 -0.0067 -0.0080 0.0064 0.0189

 

168

“In designing a national, individual animal traceability system how important is
enhancing food safety to the U.S. beef industry? ”

Table A.4.3. Enhancing Food Safety Estimates

 

 

Parameter Estimate Standard Error
Intercept 1 .83 86* * * 0.4632
Age -0.0017 0.0051
CB -0.4302 0.2759
NP 0.0314 0.2727
NW -0.1327 0.2809
SE 0.0387 0.2905
SP -0.1502 0.2883
Member -0.0072 0.0432
Education -0.03 90 0.03 51
Years 0.0077 0.0407
Eprears 0.0136 0.0264
Income -0.0112 0.0311
OIncome 0.0145 0.0361
Labor -0.0503 0.0574
OFeed -0.0622 0.0517
Cows 0.0431 0.0365
Auctions -0. 1025 0.1082
RegNAIS 0.3008*** 0.1034
Limit 2 0.3188*** 0.0599
Limit 3 0.9589*** 0.0825
Limit 4 1.7308*** 0.0926

 

- *** 1% significance level, ** 5% significance level, "‘ 10% significance level

Table A.4.4. ME to Enhancing Food Safety

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
RegNAIS -0.0366 -0.0191 -0.0417 -0.0174 0.1148

 

169

“In designing a national, individual animal traceability system how important is
improving marketability to the U.S. beef industry? ”

Table A.4.5. Improving Marketability Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.8942* * "‘ 0.4581
Age -0.0124** 0.0050
CB -0.2591 0.2664
NP 0.0557 0.2630
NW 0.0046 0.2717
SE 0.0631 0.2798
SP 0.0768 0.2804
Member 0.0449 0.0430
Education 0.0292 0.0346
Years 0.0541 0.0395
Eprears 0.0040 0.0261
Income -0.0148 0.0306
OIncome -0.0039 0.0358
Labor -0.0374 0.0562
OFeed -0.0259 0.0513
Cows 0.0273 0.0362
Auctions -0.l943* 0.1062
RegNAIS 0.2276“ 0.1012
Limit 2 0.3355*** 0.0617
Limit 3 1.0084*** 0.0842
Limit 4 2.0225*** 0.0970

 

. *** 1% significance level, ** 5% significance level, * 10% significance level

Table A.4.6. ME to Improving Marketability

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
Age 0.0015 0.0008 0.0018 0.0002 -0.0043
Auctions 0.0236 0.0130 0.0276 0.0038 -0.0680
RegNAIS -0.0277 -0.0152 -0.0324 -0.0044 0.0797

 

170

“In designing a national, individual animal traceability system how important is
improving on-farm management to the U.S. beef industry? ”

Table A.4.7. Improving On-Farm Management Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.5890*** 0.4420
Age -0.0070 0.0049
CB 0.1508 0.2578
NP 0.2178 0.2527
NW -0.0293 0.2622
SE 0.2583 0.2702
SP 0.2735 0.2700
Member 0.01 15 0.0422
Education —0.0282 0.0345
Years 0.0188 0.0388
Eprears -0.0148 0.0258
Income -0.0326 0.0299
OIncome -0.0062 0.0353
Labor -0.0408 0.0553
OF eed -0.0583 0.0499
Cows 0.0572* 0.0350
Auctions -0.0289 0. 1043
RegNAIS 0.131 1 0.0993
Limit 2 0.3919*** 0.0534
Limit 3 l.0597*** 0.0721
Limit 4 1.8547*** 0.0844

 

. *** 1% significance level, ** 5% significance level, * 10% significance level

Table A.4.8. ME to Improving On-Farm Management

 

 

 

ME
Entirely Very
Parameter Unimportant Unimportant Neutral Important Important
Cows -0.0109 -0.0052 -0.0061 0.0036 0.0185

 

171

“Implementing individual animal traceability systems is more cost effective for larger
cow-calf operations. ”

Table A.4.9. Economies of Scale Estimates

 

 

Parameter Estimate Standard Error
Intercept 1.0683M 0.4375
Age 0.0065 0.0048
CB -0.3830 0.2627
NP -0.2211 0.2581
NW -0.3110 0.2672
SE -0.0601 0.2747
SP -0.2197 0.2734
Member 00005 0.0409
Education 0.0232 0.0334
Years -0.0052 0.03 83
Eprears -0.0213 0.0250
Income 0.0259 0.0293
OIncome 0.0609* 0.0345
Labor -0.0045 0.0548
OF eed 0.0683 0.0491
Cows -0.0641* 0.0345
Auctions -0.0042 0.1015
RegNAIS -0.0279 0.0967
Limit 2 0.5780*** 0.0631
Limit 3 l .43 83* * * 0.0798
Limit 4 2.2579*** 0.0924

 

° * 1% significance level, ** 5% significance level, *** 10% significance level

Table A.4.10. ME to Economies of Scale

 

 

 

ME
Strongly Strongly
Parameter Disagree Disagree Neutral Agree Agree
OIncome -0.0104 -0.0079 -0.0052 0.0080 0.0155
Cows 0.0110 0.0083 0.0055 -0.0084 -0.0163

 

172

“Implementing individual animal traceability results in more liability for cow-calf
producers than cattle owners at other stages of production. ”

Table A.4.ll. Shifts in Liability Estimates

 

 

Parameter Estimate Standard Error
Intercept 0.9697" 0.4443
Age -0.0045 0.0049
CB -0.0114 0.2633
NP 0.1679 0.2593
NW 0.0483 0.2683
SE 0.3616 0.2759
SP 0.1634 0.2751
Member -0.0226 0.041 5
Education 0.0264 0.0343
Years 0.0661 * 0.0386
Eprears -0.0244 0.0254
Income 0.0585* 0.0299
OIncome -0.0016 0.0352
Labor -0.0492 0.0561
OFeed 0.0482 0.0504
Cows 0.0732“ 0.0350
Auctions 0.1275 0.1027
RegNAIS -0.2539** 0.0989
Limit 2 0.7544*** 0.0892
Limit 3 1.4416*** 0.1003
Limit 4 2.3166*** 0.1085

 

- *** 1% significance level, ** 5% significance level, * 10% significance level

Table A.4.12. ME to Shifts in Liability

 

 

 

ME
Strongly Strongly
Parameter Disagree Disagree Neutral Agree Agree
Years -0.0059 -0.0098 -0.0085 0.0020 0.0222
Income —0.0052 -0.0087 -0.0076 0.0018 0.0197
Cows -0.0065 -0.0109 -0.0095 0.0022 0.0246
RegNAIS 0.0225 0.0378 0.0329 -0.0078 -0.0854

 

173

“Implementing individual animal traceability systems as a mandated system is
exaggerated in need. ”

Table A.4.13. Mandatory Traceability Exaggeration Estimates

 

 

Parameter Estimate Standard Error
Intercept 0.7849* 0.4364
Age -0.0018 0.0049
CB -0.0774 0.2599
NP 0.0481 0.2550
NW 0.1256 0.2646
SE 0.1857 0.2715
SP 0.1464 0.2716
Member -0.0413 0.0413
Education 0.0164 0.0340
Years 0.0976* * 0.0385
Eprears -0.0134 0.0252
Income -0.0005 0.0296
OIncome -0.0446 0.0349
Labor -0.0202 0.0553
OFeed 0.0804 0.0497
Cows 0.0066 0.0346
Auctions 0.1565 0.1025
RegNAIS -0.2841* * * 0.0981
Limit 2 0.6262* * * 0.0672
Limit 3 1.5078*** 0.0835
Limit 4 2.1969*** 0.0932

 

. *** 1% significance level, ** 5% significance level, * 10% significance level

Table A.4.14. ME to Mandatory Traceability Exaggeration

 

 

 

Strongly Strongly
Parameter Disagree Disagree Neutral Agree Agree
Years -0.0150 -0.0134 -0.0091 0.0094 0.0281
RegNAIS 0.0435 0.0389 0.0266 -0.0272 -0.0818

 

174

APPENDIX 5: SAS, STATA, AND LIMDEP CODE
5.1 SAS
*/F actor Analysis SAS Code*/

PROC IMPORT OUT: WORK.rawdata

DATAFILE= "EzEstimation Data Set 05.04.08.xls"

DBMS=EXCEL REPLACE; SHEET="Non CE Est Data"; GETNAMES=YES;
MIXED=NO; SCANTEXT=YES; USEDATE=YES; SCANTIME=YES;
RANGE="A1:CE656";RUN;

title 'proc means data=rawdata;run;';
proc means data=rawdata;run;

/**********************************************************************/

/*FIRST CONSIDER ALL INDICATORS IN Q17, 18, and 19*/

titlel 'Conduct a principal component analysis (Q17-19)‘;

proc factor data=rawdata method=principal score

outstat=factor_a rotate=varimax mineigen=1;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals asce asliabi ascool asmanex;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_a out=scores_factor_a;run;

/*compare with forcing a 4th factor*/

/*note this results in the 3rd and 4th factors having only 2 indicators each*/

title] 'Conduct a principal component analysis (Q17-19)‘;

proc factor data=rawdata method=principal score

outstat=factor_a4 rotate=varimax nfactors=4;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals asce asliabi ascool asmanex;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_a4 out=scores_factor_a4;run;

/*compare with forcing a 5th factor*/

/*note this results in the 3rd, 4th, & 5th factors having only 2 indicators each*/
titlel 'Conduct a principal component analysis (Q17-19)’;

proc factor data=rawdata method=principal score

outstat=factor_a5 rotate=varimax nfactors=5;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty

cricost criconf crirelia criliabi crifride crigoals asce asliabi ascool asmanex;run;

175

title2 'compute principal component scores';
proc score data=rawdata score=factor_a5 out=scores_factor_a5;run;

/*NOW CONSIDER ONLY INDICATORS IN Q17 & Q18*/

titlel 'Conduct a principal component analysis (Q17-18)‘;

proc factor data=rawdata method=principal score

outstat=factor_b rotate=varimax mineigen=1;

var ioidis ioicons ioimarkt ioimcfin ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_b out=scores_factor_b;run;

/ * compare with forcing a 3rd factor*/

titlel 'Conduct a principal component analysis (Q17-18)‘;

proc factor data=rawdata method=principal score
outstat=factor_b3 rotate=varimax nfactors=3;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty
cricost criconf crirelia criliabi crifride crigoals;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_b3 out=scores_factor_b3;run;

/*compare with forcing 4 factors*/

titlel 'Conduct a principal component analysis (Q17-18)‘;

proc factor data=rawdata method=principal score
outstat=factor_b4 rotate=varimax nfactors=4;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty
cricost criconf crirelia criliabi crifride crigoals;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_b4 out=scores_factor_b4;run;

/*AS A SIDE EVALUATION, A FACTOR ANAYSIS INCORPORATIN G THE Q21
BENEFIT/COST VALUES WITH Q17-19 WAS PERFORMED:*/

titlel ’Conduct a principal component analysis (Q17-19 & 21)’;

proc factor data=rawdata method=principal score

outstat=factor_c rotate=varimax mineigen=l;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty

cricost criconf crirelia criliabi crifride crigoals asce asliabi ascool asmanex benret
benproc benfdlp benccp costret costproc costfdlp costccp;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_c out=scores_factor_c;run;

176

/*compare with forcing only 5 factors*/

titlel 'Conduct a principal component analysis (Q17-19 & 21)’;

proc factor data=rawdata method=principal score

outstat=factor_c5 rotate=varimax nfactors=5;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals asce asliabi ascool asmanex benret benproc benfdlp benccp
costret costproc costfdlp costccp;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_c5 out=scores_factor_cS;run;

/*compare with forcing 7 factors*/

titlel 'Conduct a principal component analysis (Q17-l9 & 21)’;

proc factor data=rawdata method=principal score

outstat=factor_c7 rotate=varimax nfactors=7;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals asce asliabi ascool asmanex benret benproc benfdlp benccp
costret costproc costfdlp costccp;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_c7 out=scores_factor_c7;run;

/*compare to only including the cow-calf portion of Q21 */

titlel 'Conduct a principal component analysis (Q17-19 & 21 cow-calf)’;

proc factor data=rawdata method=principal score

outstat=factor_c2 rotate=varimax mineigen=l;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty cricost criconf crirelia
criliabi crifride crigoals asce asliabi ascool asmanex benccpcostccp;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_02 out=scores_factor_c2;run;

/*NOW CONSIDER ONLY INDICATORS IN Q17*/

titlel 'Conduct a principal component analysis (Q17)';

proc factor data=rawdata method=principal score

outstat=factor_b rotate=varimax mineigen=l;

var ioidis ioicons ioimarkt ioimcfm ioiafm ioimngt ioischn ioifsfty;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_b out=scores_factor_b;run;

/*NOW CONSIDER ONLY INDICATORS IN Q18*/
titlel 'Conduct a principal component analysis (18)';
proc factor data=rawdata method=principa1 score
outstat=factor_b rotate=varimax mineigen=l;

var cricost criconf crirelia criliabi crifride crigoals;run;

177

title2 'compute principal component scores';
proc score data=rawdata score=factor_b out=scores_factor_b;run;

/*NOW CONSIDER ONLY INDICATORS IN Q19*/
titlel 'Conduct a principal component analysis (18)';
proc factor data=rawdata method=principa1 score
outstat=factor_b rotate=varimax mineigen=l;

var asce asliabi ascoolasmanex;run;

title2 'compute principal component scores';
proc score data=rawdata score=factor_b out=scores_factor_b;run;

/******************************************************************#**/
/*********************************************************************/

*/Tobit and Probit Estimation SAS Code*/

PROC IMPORT OUT= WORK.rawdata

DATAFILE= "EzEstimation Data Set 05.04.08.xls"

DBMS=EXCEL REPLACE; SHEET="Non CE Est Data"; GETNAMES=YES;
MIXED=NO; SCANTEXT=YES; USEDATE=YES; SCANTIME=YES;
RANGE="A1:CY656"; RUN;

title 'proc means data=rawdata;run;';
proc means data=rawdata;run;

/******************************#**************************************/

*Binary Probit for Evaluating Question 22 (regnais)*/
*(22)Are your operation's premise(s) currently registered with USDA in the NAIS
(National Animal Identification System)? (Yes or No)*/

title 'Estimation of regnais‘;

PROC QLIM DATA=Rawdata outest=outest;

MODEL regnais = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla dumpetid durnenid dumbrdid dumtatid dumbrid dumrfid
dumnoid mannais /discrete (dist=norma1);output out=outMEregnais marginal;

nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEregnais';
proc means data=outMEregnais;run;

/*********************************************************************/

178

*Tobit for Evaluating Question 24 (fnais)*/

*(24)If the NAIS system remains a system of voluntary participation, by December 31,
2008 what do you predict will be (please circle your answers):*/

*(a)The percentage of U.S. cow-calf operations with premises registered in NAIS.*/

title 'Joint Estimation of fnais & regnais';

PROC QLIM DATA=Rawdata outest=outestfnaisreg;

MODEL regnais = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla dumpetid dumenid dumbrdid dumtatid dumbrid dumrfid
dumnoid mannais /discrete (dist=normal);

MODEL fnais = age countfo yrsr expyrsr bctc dumla dumrfid regnais mannais cretmccp
cprocmccp cfdlpmccp/discrete (dist=normal);endogenous fnais ~ censored(lb=0 ub=100);
nloptions maxiter=500;RUN;

*Since there is no endogeneity with regnais*/

title 'Estimation of fnais';

PROC QLIM DATA=Rawdata outest=outestfnais;

MODEL fnais = age countfo yrsr expyrsr bctc dumla dumrfid regnais mannais cretmccp
cprocmccp cfdlpmccp/discrete (dist=normal);endogenous fnais ~ censored(lb=0 ub=100);
output out=outMEfnais marginalmloptions maxiter=500;RUN;

/*now Show summary of marginals (across individuals)*/
titlel 'marginals from outMEfnais';
proc means data=outMEfnais;run;

/*********************************************************************/

*(b)The percentage of U.S. cow-calf operations with premises registered in NAIS AND
using NAIS approved RFID animal identification devices on the percentage of cattle
leaving their premises. (fnaisrfid)*/

title 'Joint Estimation of fnaisrfid & regnais';

PROC QLIM DATA=Rawdata outest=outest;

MODEL regnais = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla dumpetid dumenid dumbrdid dumtatid dumbrid dumrfid
dumnoid mannais /discrete (dist=normal);

MODEL fnaisrfid = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla dumpetid dumenid dumbrdid dumtatid dumbrid dumrfid
dumnoid regnais mannais/discrete (dist=normal);

endogenous fnaisrfid ~ censored(lb=0 ub=100); nloptions maxiter=500;RUN;

179

*Since there is no endogeneity with regnais*/

title ’Estimation of fnaisrfid';

PROC QLIM DATA=Rawdata outest=outest;

MODEL fnaisrfid = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla dumpetid dumenid dumbrdid dumtatid dumbrid dumrfid
dumnoid regnais mannais/discrete (dist=normal);

endogenous fnaisrfid ~ censored(lb=0 ub=100);output out=outMEfnaisrfid marginal;
nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEfnaisrfid';
proc means data=outMEfnaisrfid;run;

/*********************************************************************/

*Ordered Probit for Evaluating Questions 17-19*/

*(17)In designing a national, individual animal traceability system how important are the
following issues in the U.S. beef industry (please circle your answers where 1 = Entirely
Unimportant, 2 = Unimportant, 3 = Neutral, 4 = Important, 5 = Very Important):*/

title 'Estimation of ioidis';

PROC QLIM DATA=Rawdata outest=outestdis covb;

MODEL ioidis = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEdis marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEdis';
proc means data=outMEdis;run;

/*********************************************************************/

title 'Estimation of ioicons';

PROC QLIM DATA=Rawdata outest=outestcons covb;

MODEL ioicons = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEcons marginal;nloptions maxiter=500;RUN;

/* now show summary of marginals (across individuals)*/
titlel 'marginals from outMEcons';
proc means data=outMEcons;run;

/*********************************************************************/

180

title 'Estimation of ioimarkt';

PROC QLIM DATA=Rawdata outest=outestmarkt covb;

MODEL ioimarkt = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr
income oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEmarkt marginalmloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEmarkt';
proc means data=outMEmarkt;run;

/*********************************************************************/

title 'Estimation of ioimcfm';

PROC QLIM DATA=Rawdata outest=outestmcfm covb;

MODEL ioimcfm = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr
income oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEmcfm marginalmloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEmcfm';
proc means data=outMEmcfm;run;

/*********************************************************************/

title 'Estimation of ioiafm';

PROC QLIM DATA=Rawdata outest=outestafm covb;

MODEL ioiafm = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEafm marginalmloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEafm';
proc means data=outMEafm;run;

/*********************************************************************/

title 'Estimation of ioimngt';

PROC QLIM DATA=Rawdata outest=outestmgt covb;

MODEL ioimngt = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEmgt marginalmloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEmngt';

proc means data=outMEmgt;run;
/*********************************************************************/

181

title 'Estimation of ioischn';

PROC QLIM DATA=Rawdata outest=outestschn covb;

MODEL ioischn = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEschn marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEschn';
proc means data=outMEschn;run;

/********************#************************************************/

title 'Estimation of ioifsfty';

PROC QLIM DATA=Rawdata outest=outestfsfty covb;

MODEL ioifsfty = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEfsfiy marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEfsfty';
proc means data=outMEfsfty;run;

/*********************************************************************/

*(18)In designing a national, individual animal traceability system how concerned are
you regarding the following issues in the U.S. beef industry (where 1 = Entirely
Unconcerned, 2 = Unconcerned, 3 = Neutral, 4 = Concerned, 5 = Very Concemed):*/

title 'Estimation of cricost';

PROC QLIM DATA=Rawdata outest=outestcost covb;

MODEL cricost = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEcost marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*l
titlel 'marginals from outMEcost';
proc means data=outMEcost;run;

/*********************************************************************/

title 'Estimation of criconf;

PROC QLIM DATA=Rawdata outest=outestconf covb;

MODEL criconf = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEconf marginal;nloptions maxiter=500;RUN;

182

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEconf;
proc means data=outMEconf;run;

/**********************************************************************/

title 'Estimation of crirelia';

PROC QLIM DATA=Rawdata outest=outestrelia covb;

MODEL crirelia = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMErelia marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMErelia';
proc means data=outMErelia;run;

**********************************************************************/

title 'Estimation of criliabi';

PROC QLIM DATA=Rawdata outest=outestliabi covb;

MODEL criliabi = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEliabi marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEliabi';
proc means data=outMEliabi;run;

*****#****************************************************************/

title 'Estimation of crifride';

PROC QLIM DATA=Rawdata outest=outestfride covb;

MODEL crifride = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEfride marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*I
titlel 'marginals from outMEfride';
proc means data=outMEfride;run;

**********************************************************************/

183

title 'Estimation of crigoals';

PROC QLIM DATA=Rawdata outest=outestgoals covb;

MODEL crigoals = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEgoals marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEgoals';
proc means data=outMEgoals;run;

/*********************************************************************/

*(19)Please circle your level of agreement with each of the following
statements (Where l=Strongly Disagree, 2=Disagree, 3=Neutral, 4=Agree,
5=Strongly Agree): Implementing individual animal traceability sytems*/

title 'is more cost effective for larger cow-calf operations';

PROC QLIM DATA=Rawdata outest=outestsca1e covb;

MODEL asce = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEscale marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEscale';
proc means data=outMEscale;run;

/*********************************************************************/

title 'results in more liability for cow-calf producers';

PROC QLIM DATA=Rawdata outest=outestrliability covb;

MODEL asliabi = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMErliabi marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMErliabi';
proc means data=outMErliabi;run;

/*********************************************************************/

title 'is unnecessary if COOL';

PROC QLIM DATA=Rawdata outest=outestcool covb;

MODEL ascool = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr income
oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEcool marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/

184

titlel 'marginals from outMEcooI';
proc means data=outMEcool;run;

/*********************************************************************/

title 'as a mandated system is exaggerated in need';

PROC QLIM DATA=Rawdata outest=outestasmanex covb;

MODEL asmanex = age erscb ersnp ersnwa erssea erssp countfo edu yrsr expyrsr
income oincome labor ofeed bctc dumla regnais /discrete (dist=normal);

output out=outMEmanex marginal;nloptions maxiter=500;RUN;

/*now show summary of marginals (across individuals)*/
titlel 'marginals from outMEmanex';
proc means data=outMEmanex;run;

/**********************************************************************/
/**********************************************************************/

/*Panel Data Development and Effects Coding SAS Code*/

PROC IMPORT OUT= WORK.rawdata

DATAFILE= "EzEstimation Data Set 05.04.08.xls"

DBMS=EXCEL REPLACE; SHEET="SASImport"; GETNAMES=YES;
MIXED=NO; SCANTEXT=YES; USEDATE=YES; SCANTIME=YES;
RANGE="A1:ER656"; RUN;

/* create a panel data set of 655*4* 3 total observations since CE has 4 scenarios and 3
options per scenario*/

premium/discount denoted by PD
managing entity denoted by ME
additional information denoted by AI

data Rawdata_Panel;
set Rawdata;

retain SurveyNumb 0;
SurveyNumb+l ;

/"‘T1 - denotes lst observation in 4 observation panel*/
T=1;T1=1;T2=0;T3=0;T4=0;

decision = S 1 ;PD = SlPDa; ME=SIMEa;AI=SlAIa;Adum = 1; Bdum = 0;Cdum = 0;
output;

185

decision = Sl;PD = SlPDb; ME=SIMEb;AI=SIAIb;Adum = 0; Bdum = 1;Cdum = 0;

output;

decision = S 1 ;PD = SlPDc; ME=SIMEc;AI=SlAIc;Adum = 0; Bdum = 0;Cdum = 1;

output;

/*T2 - denotes 2nd observation in 4 observation panel*/
T=2;T1=0;T2=1;T3=0;T4=0;

decision = SZ;PD = S2PDa; ME=SZMEa;AI=SZAIa;Adum = 1; Bdum = 0;Cdum = 0;

output;

decision = S2;PD = SZPDb; ME=S2MEb;AI=SZAIb;Adum = 0; Bdum = 1;Cdum = 0;

output;

decision = 82;PD = SZPDc; ME=S2MEc;AI=SZAIc;Adum = 0; Bdum = 0;Cdum = 1;

output;

/*T3 - denotes 3rd observation in 4 observation panel*l
T=3;T1=0;T2=0;T3=1;T4=0;

decision = S3;PD = S3PDa; ME=S3MEa;AI=S3AIa;Adum = l; Bdum = 0;Cdum = 0;

output;

decision = S3;PD = S3PDb; ME=S3MEb;AI=S3AIb;Adum = 0; Bdum = 1;Cdum = 0;

output;

decision = S3;PD = S3PDc;ME=S3MEc;AI=S3AIc;Adum = 0; Bdum = 0;Cdum = 1;

output;

/*T4 - denotes 4th observation in 4 observation panel*/
T=4;T1=0;T2=0;T3=0;T4=1;

decision = S4;PD = S4PDa; ME=S4MEa;AI=S4AIa;Adum = 1; Bdum = 0;Cdum = 0;

output;

decision = S4;PD = S4PDb; ME=S4MEb;AI=S4AIb;Adum = 0; Bdum = 1;Cdum = 0

output;

9

decision = S4;PD = S4PDc; ME=S4MEc;AI=S4AIc;Adum = 0; Bdum = 0;Cdum = 1;

output;
run;

data Rawdata_Pane12;set Rawdata_Panel;

186

/*note that 7860=655*3*4*/
modelObsID=l; do modeIObsID=l to 7860 by 3; end;

/* create a variable identifying each unique combination of individuals and CEs*/
a=RoundLn_/3);b=lag2(a);if _n_<4 then indCEobs=l ;else indCEobs=b+1;

/*change format of decision variable to Two Os and One 1 for each scenario included*/
if decision=l & Adum=l then dchse=l;
if decision=l & Bdum=1 then dchse=0;
if decision=l & Cdum=l then dchse=0;

if decision=2 & Adum=1 then dchse=0;
if decision=2 & Bdum=l then dchse=l;
if decision=2 & Cdum=1 then dchse=0;

if decision=3 & Adum=l then dchse=0;
if decision=3 & Bdum=1 then dchse=0;
if decision=3 & Cdum=l then dchse=1;

/*create a "mode" variable to track 1,2,3 options in each scenario*/
if Adum=l then mode=l;
if Bdum=l then mode=2;
if Cdum=1 then mode-=3;

/"‘create 3 dummy vars for managing entity*/
if ME=1 then ME_GV=1; else ME_GV=0;
if ME=2 then ME_PN=1; else ME_PN=0;

if ME=3 then ME_PI=1; else ME_PI=0;

/*create effects coded variables rather than dummy coded variables: base of Government
Managed takes a value of -1 when applicable, note that I added the Cdum=1 condition to
give these variables a value of 0 in the 3rd (No Trace) alternative (price is only observed
variable)*/

if ME=2 then ME_PI_EC=l;else if ME=3 then ME_PI_EC=0;
else if ME=1 then ME_PI_EC=-l;
else if Cdum=1 then ME_PI_EC=0;

if ME=3 then ME_PN_EC=1;else if ME=2 then ME_PN_EC=0;
else if ME=1 then ME_PN_EC=-1;
else if Cdum=l then ME_PN_EC=0;

/*need to create interactions with effects coded variables*/

ME_PI_EC_AD=ME_PI__EC*Adum;ME_PI_EC_BD=ME_PI_EC*Bdum;
ME_PN_EC__AD=ME_PN_EC*Adum;ME_PN_EC_BD=ME_PN_EC*Bdum;

187

/*create 15 dummy vars for additional information*/

if AI=1 then AI_AV=1; else AI_AV=0;if AI=2 then AI_PP=1; else AI_PP=0;
if AI=3 then AI_PG=1; else AI_PG=0;if AI=4 then AI_HV=1; else AI_HV=0;
if AI=5 then AI_AVPP=1; else AI_AVPP=0;

if AI=6 then AI_AVPG=1; else AI_AVPG=0;

if AI=7 then AI_AVHV=1; else AI_AVHV=0;

if AI=8 then AI_PPPG=1; else AI_PPPG=0;

if AI=9 then AI_PPHV=1; else AI_PPHV=0;

if AI=10 then AI_PGHV =1; else AI_PGHV=0;

if AI=ll then AI_AVPPPG=1; else AI_AVPPPG=0;

if AI=12 then AI_AVPPHV=1; else AI_AVPPI—IV=0;

if AI=13 then AI_PPPGHV=1; else AI_PPPGHV=0;

if AI=14 then AI_AVPPPGHV=1; else AI_AVPPPGHV=0;

/* create effects coded dummies for additional information information: base of AI_AV;
takes a value of -1 when applicable, note that Adum=l and Cdum=1 condition give these
variables a value of 0 in the lst (N AIS) alternative (price and managing entity are only
observed variableand 3rd (N 0 Trace) alternative (price is only observed variable)*/

if AI=2 then AI_PP_EC=l;else if AI=1 then AI__PP_EC=-l;

else if AI=3 then AI_PP_EC=0;else if AI=4 then AI_PP_EC=0;

else if AI=5 then AI_PP_EC=0;else if AI=6 then AI_PP_EC=0;

else if AI=7 then AI_PP_EC=0;else if AI=8 then AI_PP_EC=0;

else if AI=9 then AI_PP_EC=0;else if AI=10 then AI_PP_EC=0;

else if AI=11 then AI_PP_EC=0;else if AI=12 then AI_PP_EC=0;

else if AI=13 then AI_PP_EC=0;else if AI=14 then AI_PP_EC=0;

else if Adum=l then AI_PP_EC=0;else if Cdum=l then AI_PP_EC=0;

if AI=3 then AI_PG_EC=1;else if AI=1 then AI_PG_EC=-l;

else if AI=2 then AI_PG_EC=0;else if AI=4 then AI_PG_EC=0;

else if AI=5 then AI_PG_EC=0;else if AI=6 then AI_PG_EC=0;

else if AI=7 then AI_PG_EC=0;else if AI=8 then AI_PG_EC=0;

else if AI=9 then AI_PG_EC=0;else if AI=10 then AI_PG_EC=0;

else if AI=11 then AI_PG_EC=0;else if AI=12 then AI_PG_EC=0;

else if AI=13 then AI_PG_EC=0;else if AI=14 then AI_PG_EC=0;

else if Adum=l then AI_PG_EC=0;else if Cdum=1 then AI_PG_EC=0;

if AI=4 then AI_HV_EC=l;else if AI=1 then AI_HV_EC=-1;

else if AI=2 then AI_HV_EC=0;else if AI=3 then AI_HV_EC=0;

else if AI=5 then AI_HV_EC=0;else if AI=6 then AI_HV_EC=0;

else if AI=7 then AI_HV_EC=0;else if AI=8 then AI_HV_EC=0;

else if AI=9 then AI_HV_EC=0;else if AI=10 then AI_HV_EC=0;

else if AI=11 then AI_HV_EC=0;else if AI=12 then AI_HV_EC=0;

else if AI=13 then AI_HV_EC=0;else if AI=14 then AI_HV_EC=0;

else if Adum=l then AI_HV_EC=0;else if Cdum=l then AI_HV_EC=0;

188

if AI=5 then AI_AVPP_EC=l;else if AI=1 then AIyAVPP_EC=-l;

else if AI= 2 then AI_ AVPP_ EC=0; else if AI=3 then AI __AVPP EC=0;

else if AI= 4 then AI _AVPP _EC=0; else if AI=6 then AI _AVPP _EC —0;

else if AI= 7 then AI_AVPP_EC=0; else if AI= 8 then AI_AVPP_EC=0;

else if AI=9 then AI __AVPP EC=0; else if AI=10 then AI _AVPP _EC=0;

else if AI=11 then AI _AVPP _EC— ‘0; else if AI=12 then AI _AVPP _EC=0;

else if AI=13 then AI_AVPP_EC=0; else if AI=14 then AI_AVPP_EC— =0

else if Adum=l then AI_AVPP_EC=0;else if Cdum=1 then AI_AVPP_EC=0;

if AI=6 then AI_AVPG_EC=1;else if AI=l then AI_AVPG_EC=-1;

else if AI=2 then AI_AVPG_EC=0;eIse if AI=3 then AI_AVPG_EC=0;

else if AI=4 then AI_AVPG_EC=0;eIse if AI=5 then AI_AVPG_EC=0;

else if AI=7 then AI_AVPG_EC=0;eIse if AI=8 then AI_AVPG_EC=0;

else if AI=9 then AI_AVPG_EC=0;eIse if AI=10 then AI_AVPG_EC=0;

else if AI=11 then AI_AVPG_EC=0;eIse if AI=12 then AI_AVPG_EC=0;

else if AI=l3 then AI_AVPG_EC=0;eIse if AI=l4 then AI_AVPG_EC=0;

else if Adum=l then AI_AVPG_EC=0;eIse if Cdum=1 then AI_AVPG_EC=0;

if AI=7 then AI_AVHV_EC=1;else if AI=1 then AI_AVHV_EC=-l;

else if AI=2 then AI_AVHV_EC=0;else if AI=3 then AI_AVHV_EC=0;

else if AI=4 then AI_AVHV_EC=0;else if AI=5 then AI_AVHV_EC=0;

else if AI=6 then AI _AVHV _EC=0; else if AI=8 then AI A_VHV _EC=0;

else if AI=9 then AI _AVHV _EC— —;0 else if AI=10 then AI _AVHV _EC=0;

else if AI=11 then AI _AVHV _EC=0; else if AI=12 then AI _AVHV _=EC 0;

else if AI=13 then AI_AVHV_EC=0; else if AI=14 then AI_AVHV_EC=0;

else if Adum=l then AI_AVHV_EC=0;else if Cdum=1 then AI_AVHV_EC=0;

if AI=8 then AI_PPPG_EC=1;else if AI=1 then AI_PPPG_EC=-l;

else if AI=2 then AI_PPPG_EC=0;else if AI=3 then AI_PPPG_EC=0;

else if AI=4 then AI_PPPG_EC=0;else if AI=5 then AI_PPPG_EC=0;

else if AI=6 then AI __PPPG EC=0; else if AI= 7 then AI __PPPG EC=0;

else If A19 then AI_ PPPG_ EC -0' ,else 1f AI—10 then AI __PPPG EC= 0;

else if AI=ll then AI_PPPG_EC=0; else if AI=12 then AI_PPPG_EC=0;

else if AI=13 then AI_PPPG_EC=0;else if AI=14 then AI_PPPG_EC=0;

else if Adum=l then AI_PPPG_EC=0;else if Cdum=1 then AI_PPPG_EC=0;

if AI=9 then AI_PPHV_EC=1;eIse if AI=1 then AI_PPHV_EC=-l;

else if AI=2 then AI_PPHV_EC=0;else if AI=3 then AI_PPHV_EC=0;

else if AI=4 then AI_ PPHV _EC=0; else if AI=5 then AI _PPHV _EC=0;

else if AI=6 then AI _PPHV _EC=0; else if AI= 7 then AI__ PPHV _EC —;0

else if AI= 8 then AI _PPHV _EC=0; else if AI=10 then AI _PPHV _EC-— —;0

else if AI=ll then AI _PPHV __EC= 0; else if AI=12 then AI _PPHV_ EC=0;

else if AI=13 then AI_PPHV_EC=0; else if AI=14 then AI_PPHV_EC=0;

else if Adum=1 then AI_PPHV_EC=0;else if Cdum=1 then AI_PPHV_EC=0;

189

if AI=10 then AI_PGHV_EC=1;else if AI=1 then AI_PGHV_EC=-l;

else if AI=2 then AI_PGHV_EC=0;else if AI=3 then AI_PGHV_EC=0;

else if AI=4 then AI_PGHV_EC=0;else if AI=5 then AI_PGHV_EC=0;

else if AI=6 then AI_PGHV_EC=0;else if AI=7 then AI_PGHV_EC=0;

else if AI=8 then AI_PGHV_EC=0;else if AI=9 then AI_PGHV_EC=0;

else if AI=11 then AI_PGHV_EC=0;else if AI=12 then AI_PGHV_EC=0;

else if AI=13 then AI_PGHV_EC=0;else if AI=14 then AI_PGHV_EC=0;

else if Adum=l then AI_PGHV_EC=0;else if Cdum=1 then AI_PGHV_EC=0;

if AI=11 then AI_AVPPPG_EC=l;else if AI=1 then AI_AVPPPG_EC=-1;

else if AI=2 then AI_AVPPPG_EC=0;else if AI=3 then AI_AVPPPG_EC=0;

else if AI=4 then AI A_VPPPG_ EC=0; else if AI= 5 then AI _AVPPPG _EC=0;

else if AI= 6 then AI __AVPPPG_ EC=0; else if AI= 7 then AI _AVPPPG _EC=0;

else if AI= 8 then AI _A_VPPPG EC= 0; else if AI=9 then AI __AVPPPG EC=0;

else if AI=10 then AI_AVPPPG_EC=0; else if AI=12 then AI_AVPPPG_EC=0;

else if AI=13 then AI_AVPPPG_EC=0;else if AI=14 then AI_AVPPPG_EC=0;

else if Adum=l then AI_AVPPPG_EC=0;else if Cdum=1 then AI_AVPPPG_EC=0;

if AI=12 then AI__AVPPHV_EC=l;else if Al=l then AI_AVPPHV_EC=-1;

else if AI= 2 then AI_ AVPPHV_ EC=0; else if AI=3 then AI_ AVPPHV _EC=0;

else if AI=4 then AI_ AVPPHV _EC= 0; else If AI 5 then AI __AVPPHV_ EC= —0;

else if AI=6 then AI_AVPPHV_EC=0; else if AI= 7 then AI_AVPPHV_EC=0;

else if AI=8 then AI_AVPPHV_EC=0;else if AI=9 then AI_AVPPHV_EC=0;

else if AI=10 then AI_AVPPHV_EC=0;else if AI=11 then AI_AVPPHV_EC=0;

else if AI=13 then AI_AVPPHV_EC=0;else if AI=I4 then AI_AVPPHV_EC=0;

else if Adum=l then AI_AVPPHV_EC=0;else if Cdum=1 then AI_AVPPHV_EC=0;

if AI=13 then AI_PPPGHV_EC=1;else if AI=1 then AI_PPPGHV_EC=-1;

else if AI=2 then AI_PPPGHV_EC=0;else if AI=3 then AI_PPPGHV_EC=0;

else if AI=4 then AI_ PPPGHV _;EC=0 else if AI= 5 then AI _PPPGHV _EC=0;

else if AI=6 then AI_ PPPGHV_ EC= 0' ,else if AI= 7 then AI _PPPGHV _EC= —0;

else if AI= 8 then AI:PPPGHV_EC=0; else if AI=9 then AI_PPPGHV_EC=0;

else if AI=10 then AI_PPPGHV_EC=0;else if AI=ll then AI_PPPGHV_EC=0;

else if AI=12 then AI_PPPGHV_EC=0;else if AI=14 then AI_PPPGHV_EC=0;

else if Adum=l then AI_PPPGHV_EC=0;else if Cdum=1 then AI_PPPGHV_EC=0;

if AI=14 then AI _AVPPPGHV_ EC=1' ,else if AI=1 then AI_AVPPPGHV_EC=-1;

else if AI= 2 then AI _AV_PPPGHV EC= 0; else if AI=3 then AI _AVPPPGHV _EC=0;
else if AI=4 then AI_AVPPPGHV_EC=0; else if AI= 5 then AI_AVPPPGHV_EC=0;
else if AI=6 then AI _AVPPPGHV _EC=0; else if AI= 7 then AI _AVPPPGHV _EC=0;
else if AI=8 then AI A_VPPPGHV_ EC=0; else if AI= 9 then AI _AVPPPGHV _EC=0;
else if AI=10 then AI _AVPPPGHV_ EC= 0; else if AI=11 then AI _AVPPPGHV _EC=0;
else if AI=12 then AI_AVPPPGHV_EC=0; else if AI=13 then AI_AVPPPGHV_EC=0;
else if Adum=l then AIyAVPPPGHV_EC=0;else if Cdum=1 then
AI_AVPPPGHV_EC=0;

190

/* create effects coded dummies for lst 4 additional information info*/
/*base of AV takes a value of -1 when applicable*/

if AI=2 then AI4PP_EC=l;else if AI=1 then AI4PP_EC=-l;

else if AI=3 then AI4PP_EC=0;else if AI=4 then AI4PP_EC=0;

else if AI=5 then AI4PP_EC=0;else if AI=6 then AI4PP_EC=0;

else if AI=7 then AI4PP_EC=0;else if AI=8 then AI4PP_EC=0;

else if AI=9 then AI4PP_EC=0;else if AI=10 then AI4PP_EC=0;

else if AI=ll then AI4PP_EC=0;else if AI=12 then AI4PP_EC=0;

else if AI=l3 then AI4PP_EC=0;else if AI=14 then AI4PP_EC=0;

else if Adum=l then AI4PP_EC=0;else if Cdum=1 then AI4PP_EC=0;

if AI=3 then AI4PG_EC=1;else if AI=1 then AI4PG_EC=-l;

else if AI=2 then AI4PG_EC=0;else if AI=4 then AI4PG_EC=0;

else if AI=5 then AI4PG_EC=0;else if AI=6 then AI4PG_EC=0;

else if AI=7 then AI4PG_EC=0;else if AI=8 then AI4PG_EC=0;

else if AI=9 then AI4PG_EC=0;else if AI=10 then AI4PG_EC=0;

else if AI=11 then AI4PG_EC=0;else if AI=12 then AI4PG_EC=0;

else if AI=l3 then AI4PG_EC=0;else if AI=14 then AI4PG_EC=0;

else if Adum=l then AI4PG_EC=0;else if Cdum=1 then AI4PG_EC=0;

if AI=4 then AI4HV_EC=1;else if AI=1 then AI4HV_EC=-l;

else if AI=2 then AI4HV_EC=0;else if AI=3 then AI4HV_EC=0;

else if AI=5 then AI4HV_EC=0;else if AI=6 then AI4HV_EC=0;

else if AI=7 then AI4HV_EC=0;else if AI=8 then AI4HV_EC=0;

else if AI=9 then AI4HV_EC=0;else if AI=10 then AI4HV_EC=0;

else if AI=11 then AI4HV_EC=0;else if AI=12 then AI4HV_EC=0;

else if AI=13 then AI4HV_EC=0;else if AI=14 then AI4HV_EC=0;

else if Adum=1 then AI4HV_EC=0;else if Cdum=1 then AI4HV_EC=0;

/* create effects coded dummies for lst 10 additional information inforrnation*/
/*base of AV takes a value of -1 when applicable*/

if AI=2 then AI9PP_EC=1;else if AI=1 then AI9PP_EC=-l;

else if AI=3 then AI9PP_EC=0;else if AI=4 then AI9PP_EC=0;

else if AI=5 then AI9PP_EC=0;else if AI=6 then AI9PP_EC=0;

else if AI=7 then AI9PP_EC=0;else if AI=8 then AI9PP_EC=0;

else if AI=9 then AI9PP_EC=0;else if AI=10 then AI9PP_EC=0;

else if AI=ll then AI9PP_EC=0;else if AI=12 then AI9PP_EC=0;

else if AI=13 then AI9PP_EC=0;else if AI=14 then AI9PP_EC=0;

else if Adum=l then AI9PP_EC=0;else if Cdum=1 then AI9PP_EC=0;

191

if AI=3 then AI9PG_EC=1;else if AI=1 then AI9PG_EC=-l;

else if AI=2 then AI9PG_EC=0;else if AI=4 then AI9PG_EC=0;

else if AI=5 then AI9PG_EC=0;else if AI=6 then AI9PG_EC=0;

else if AI=7 then AI9PG_EC=0;else if AI=8 then AI9PG_EC=0;

else if AI=9 then AI9PG_EC=0;else if AI=10 then AI9PG_EC=0;

else if AI=11 then AI9PG_EC=0;else if AI=12 then AI9PG_EC=0;

else if AI=13 then AI9PG_EC=0;else if AI=14 then AI9PG_EC=0;

else if Adum=1 then AI9PG_EC=0;else if Cdum=1 then AI9PG_EC=0;

if AI=4 then AI9HV_EC=1;else if AI=1 then Al9HV_EC=-l;

else if AI=2 then AI9HV_EC=0;else if AI=3 then AI9HV_EC=0;

else if AI=5 then AI9HV_EC=0;else if AI=6 then AI9HV_EC=0;

else if AI=7 then AI9HV_EC=0;else if AI=8 then AI9HV_EC=0;

else if AI=9 then AI9HV_EC=0;else if AI=10 then AI9HV_EC=0;

else if AI=11 then AI9HV_EC=0;else if AI=12 then AI9HV_EC=0;

else if AI=13 then AI9HV_EC=0;else if AI=14 then AI9HV_EC=0;

else if Adum=1 then AI9HV_EC=0;else if Cdum=1 then AI9HV_EC=0;

if AI=5 then AI9AVPP_EC=l;else if AI=1 then AI9AVPP_EC=-l;

else if AI=2 then AI9AVPP_EC=0;else if AI=3 then AI9AVPP_EC=0;

else if AI=4 then AI9AVPP_EC=0;else if AI=6 then AI9AVPP_EC=0;

else if AI=7 then AI9AVPP_EC=0;else if AI=8 then AI9AVPP_EC=0;

else if AI=9 then AI9AVPP_EC=0;else if AI=10 then AI9AVPP_EC=0;

else if AI=ll then AI9AVPP_EC=0;else if AI=12 then AI9AVPP_EC=0;

else if AI=13 then AI9AVPP_EC=0;else if AI=14 then AI9AVPP_EC=0;

else if Adum=l then AI9AVPP_EC=0;else if Cdum=1 then AI9AVPP_EC=0;

if AI=6 then AI9AVPG_EC=l;else if AI=l then AI9AVPG_EC=-l;

else if AI=2 then AI9AVPG_EC=0;else if AI=3 then AI9AVPG_EC=0;

else if AI=4 then AI9AVPG_EC=0;else if AI=5 then AI9AVPG_EC=0;

else if AI=7 then AI9AVPG_EC=0;else if AI=8 then AI9AVPG_EC=0;

else if AI=9 then AI9AVPG_EC=0;else if AI=10 then AI9AVPG_EC=0;

else if AI=11 then AI9AVPG_EC=0;else if AI=12 then AI9AVPG_EC=0;

else if Al=l3 then AI9AVPG_EC=0;else if Al=l4 then AI9AVPG_EC=0;

else if Adum=1 then AI9AVPG_EC=0;else if Cdum=1 then AI9AVPG_EC=0;

if AI=7 then AI9AVHV_EC=l;else if AI=1 then AI9AVHV_EC=-l;

else if AI=2 then AI9AVHV_EC=0;else if AI=3 then AI9AVHV_EC=0;

else if AI=4 then AI9AVHV_EC=0;else if AI=5 then AI9AVHV_EC=0;

else if AI=6 then AI9AVHV_EC=0;else if AI=8 then AI9AVHV_EC=0;

else if AI=9 then AI9AVHV_EC=0;else if AI=10 then AI9AVHV_EC=0;

else if AI=11 then AI9AVHV_EC=0;else if AI=12 then AI9AVHV_EC=0;

else if AI=13 then AI9AVHV_EC=0;else if AI=14 then AI9AVHV_EC=0;

else if Adum=l then AI9AVHV_EC=0;else if Cdum=1 then AI9AVHV_EC=0;

192

if AI=8 then AI9PPPG_EC=1;else if AI=1 then AI9PPPG_EC=-l;

else if AI=2 then AI9PPPG_EC=0;else if AI=3 then AI9PPPG_EC=0;

else if AI=4 then AI9PPPG_EC=0;else if AI=5 then AI9PPPG_EC=0;

else if AI=6 then AI9PPPG_EC=0;else if AI=7 then AI9PPPG_EC=0;

else if AI=9 then AI9PPPG_EC=0;else if AI=10 then AI9PPPG_EC=0;

else if AI=11 then AI9PPPG_EC=0;else if AI=12 then AI9PPPG_EC=0;

else if AI=13 then AI9PPPG_EC=0;else if AI=14 then AI9PPPG_EC=0;

else if Adum=l then AI9PPPG_EC=0;else if Cdum=1 then AI9PPPG_EC=0;

if AI=9 then AI9PPHV_EC=1;else if AI=l then AI9PPHV_EC=-l;

else if AI=2 then AI9PPHV_EC=0;else if AI=3 then AI9PPHV_EC=0;

else if AI=4 then AI9PPHV_EC=0;else if AI=5 then AI9PPHV_EC=0;

else if AI=6 then AI9PPHV_EC=0;else if AI=7 then AI9PPHV_EC=0;

else if AI=8 then AI9PPHV_EC=0;else if AI=10 then AI9PPHV_EC=0;

else if AI=ll then AI9PPHV_EC=0;else if AI=12 then AI9PPHV_EC=0;

else if AI=13 then AI9PPHV_EC=0;else if AI=14 then AI9PPHV_EC=0;

else if Adum=l then AI9PPHV_EC=0;else if Cdum=1 then AI9PPHV_EC=0;

if AI=10 then AI9PGHV_EC=l;else if AI=1 then AI9PGHV_EC=-l;

else if AI=2 then AI9PGHV_EC=0;else if AI=3 then AI9PGHV_EC=0;

else if AI=4 then AI9PGHV_EC=0;else if AI=5 then AI9PGHV_EC=0;

else if AI=6 then AI9PGHV_EC=0;else if AI=7 then AI9PGHV_EC=0;

else if AI=8 then AI9PGHV_EC=0;else if AI=9 then AI9PGHV_EC=0;

else if Al=ll then AI9PGHV_EC=0;else if AI=12 then AI9PGHV_EC=0;

else if AI=l3 then AI9PGHV_EC=0;else if AI=14 then AI9PGHV_EC=0;

else if Adum=1 then AI9PGHV_EC=0;else if Cdum=1 then AI9PGHV_EC=0;

/*create dummy variable indicating discount:*/
if PD<0 then Discount=l;else Discount=0;
if PD>0 then Premium=l;else Premium=0;

PD_Disc=PD*Discount;
PD_Prem=PD*Premium;

run;
/*Create separate data set for only fully complete and balanced panal data set*/
data dataCEcomplete;set Rawdata_pane12;

if CEbalanced=0 then delete;

run;

/*311**IIUIUIHIUIHI‘*******************11‘Il‘*******ll:******************************/
/**=1!*********4!#101!*********IlUlI**************#***************************/

193

5.2 STATA

*Now Looking at Endogeneity Bias with regnais*/

dprobit regnais age erscb ersnp ersnwa erssea erssp countfo yrsr expyrsr edu income
oincome labor ofeed bctc dumla dumpetid dumenid dumbrdid dumtatid dumbrid dumrfid
dumnoid mannais

*Predicting Inverse Mills Ratio*/

predict yhat, xb

generate phi = (1/sqrt(2*_pi))*exp(-(yhat"2/2))
generate capphi = norm(yhat)

generate invmills = phi/capphi

*Trinomial Probit or Nested Probit for Evaluating Question 23 (mannais)*/
*(23)Do you believe that NAIS (as previously outlined) should be a mandatory system
requiring all U.S. cattle producers to participate? (Yes, No, or Undecided)*/

*Final Regression After Testing*/
mprobit mannais age erscb ersnp ersnwa erssea erssp countfo yrsr expyrsr income bctc
dumla bretmccp bprocmccp bfdlpmccp cretmccp cprocmccp cfdlpmccp invmills

mfx compute, predict(outcome(2)) dydx at(mean)
mfx compute, predict(outcome(1)) dydx at(mean)
mfx compute, predict(outcome(0)) dydx at(mean)

/***IIUIUIHII***************************************************************/
/**********************************************************************/

194

5.3 LIMDEP

$LIMDEP Code for CE Estimation$
RESET
READ;FILE="E:\LIMDEP__Import.xls"

$Summary Statistics$
DSTAT;Rhs=DECUSE,PD,PD_prem,PD_disc,Cdum,Bdum,ME_PI_EC,ME_PN_EC,
AI4PP_EC,AI4PG_EC,AI4HV_EC,bctc,countfo,yrsr,dum1a,indv_id,group_id,no_ot_id,
erscb,ersnp,ersnwa,erssea,erssp,AI9PP_EC,AI9PG_EC,AI9HV_EC,AI9AVPP_,
AI9AVPG_,AI9AVHV_,AI9PPPG_,AI9PPHV_,AI9PGHV_,AI_PP_EC,AI_PG_EC,
AI_HV_EC,AI_AVPP_,AI_AVPG_,AI_AVHV_,AI_PPPG_,AI_PPHV_,AI_PGHV_,
AI_AVPPP,AI_AVPPH,AI_PPPGH,AI_FL_EC,BD_erscb,CD_erscb,BD_ersnp,
CD_ersnp,BD_ersnw,CD_ersnw,BD_ersse,CD_ersse,BD_erssp,CD_erssp$

$Important DECUSE in this code has a mean of 0.33333, so CE models work$
$Creating BDUM and CDUM interactions with Demos

CREA;Fill;BD_age=Bdum*age$
CREA;Fill;CD_age=Cdum*age$
CREA;Fill;BD_bctc=Bdum*bctc$
CREA;Fill;CD_bctc=Cdum*bctc$
CREA;Fill;BD_dumla=Bdum*dumla$
CREA;Fill;CD_dumla=Cdum*dumla$
CREA;Fill;BD_countfo=Bdum*countfo$
CREA;Fill;CD_countfo=Cdum*countfo$
CREA;Fill;BD_erscb=Bdum*erscb$
CREA;Fill;CD_erscb=Cdum*erscb$
CREAgFill;BD_ersnc=Bdum*ersnc$
CREA;Fill;CD_ersnc=Cdum*ersnc$
CREA;F i11;BD_ersnp=Bdum*ersnp$
CREA;FiIl;CD__ersnp=Cdum*ersnp$
CREA;Fill;BD_ersnwa=Bdum*ersnwa$
CREA;Fi1l;CD_ersnwa=Cdum*ersnwa$
CREA;Fill;BD_erssea=Bdum*erssea$
CREA;Fill;CD_erssea=Cdum*erssea$
CREA;F i11;BD_erssp=Bdum*erssp$
CREA;Fi11;CD__erssp=Cdum*erssp$
CREA;Fill;BD_yrsr=Bdum*yrsr$
CREA;Fill;CD_yrsr=Cdum*yrsr$
CREA;Fill;BD_indid=Bdum*indv_id$
CREA;Fill;CD_indid=Cdum*indv__id$
CREA;Fill;BD_grpid=Bdum*group_id$

195

CREA;Fi11;CD_grpid=Cdum*group_id$
CREA;F ill ;BD_notid=Bdum* no_ot_id$
CREA;Fill;CD_notid=Cdum*no_ot_id$

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$Preferred Models
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

$MNL

$MNL (PD & Demos & ERS Regions & 13/14 AI)
NLOGIT;th=DECUSE;Choices=1,2,3;
Rhs=PD,Cdum,Bdum,ME_PI_EC,ME_PN_EC,
AI_PP_EC,AI_PG_EC,AI_HV_EC,AI_AVPP_,AI__AVPG_,AI_AVHV_,AI_PPPG_,
AI_PPHV_,AI_PGHV_,AI_AVPPP,AI_AVPPH,AI_PPPGH,AI_FL_EC,BD_bctc,
CD__bctc,BD_dumla,CD_dumla,BD_count,CD_count,BD_yrsr,CD_yrsr,BD_indid,
CD_indid,BD_grpid,CD_grpid,BD_notid,CD_notid,BD_erscb,CD_erscb,BD_ersnp,
CD_ersnp,BD_ersnw,CD_ersnw,BD_ersse,CD__ersse,BD_erssp,CD_erssp;
Pds=4;PrintVC;Effects:;Means:;Matrix;Crosstab;Prob=prMN5$

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$RPL Not Correlated

$RPL MODEL (PD & Demos & 13/14 AI):RANDOM PARMS ARE NOT
CORRELATED

NLOGIT;th=DECUSE;Choices=1,2,3;
Rhs=PD_prem,PD_disc,Cdum,Bdum,ME_PI_EC,ME__PN_EC,AI_PP_EC,AI_PG_EC,
AI_HV_EC,AI_AVPP_,AI_AVPG_,AI_AVHV_,AI_PPPG_,AI_PPHV_,AI_PGHV__,
AI_AVPPP,AI_AVPPH,AI_PPPGH,AI_FL_EC,BD_bctc,CD_bctc,BD_dum1a,
CD_dumla,BD_count,CD_count,BD_yrsr,CD_yrsr,BD_indid,CD_indid,BD_grpid,
CD_grpid,BD_notid,CD_notid;

RPL;Pts=250;Parameters;

Fcn=Cdum(N),Bdum(N);
PrintVC;Effects:;halton;Means;Maxit=100;Pds=4;Matrix;Crosstab;Prob=erP5nc$

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

196

$RPL Correlated

$RPL MODEL (PD & Demos & ERS Regions & 13/14 AI):RANDOM PARMS ARE
CORRELATED W/ DEMOS

NLOGIT;th=DECUSE;Choices=1,2,3;
Rhs=PD,Cdum,Bdum,ME_PI_EC,ME_PN_EC,AI_PP_EC,AI_PG_EC,AI_HV_EC,
AI_AVPP_,AI_AVPG_,AI_AVHV_,AI_PPPG__,AI_PPHV_,AI_PGHV_,AI_AVPPP,
AI_AVPPH,AI_PPPGH,AI_FL_EC,BD_bctc,CD_bctc,BD_dumla,CD_dumla,BD_count,
CD__count,BD_yrsr,CD_yrsr,BD_indid,CD_indid,BD__grpid,CD_grpid,BD_notid,
CD_notid,BD_erscb,CD_erscb,BD_ersnp,CD_ersnp,BD_ersnw,CD_ersnw,BD_ersse,
CD_ersse,BD_erssp,CD_erssp;

RPL;Pts=250;Cor;Parameters;

F cn=Cdum(N ),Bdum(N);
PrintVC;Effects:;halton;Means;Maxit=100;Pds=4;Matrix;Crosstab;Prob=erP5c$

$$$$$$8$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$LCM

$LCM MODEL (PD & Demos & 9/10 AI): 3 SEGMENTS:

$NO SEGEMENT EXPLAININ G VARIABLES INCLUDED
NLOGIT;th=DECUSE;Choices=l ,2,3;
Rhs=PD,Cdum,Bdum,ME_PI_EC,ME_PN_EC,AI9PP_EC,AI9PG_EC,AI9HV_EC,
AI9AVPP_,AI9AVPG_,AI9AVHV_,A19PPPG_,AI9PPHV_,AI9PGHV_;
LCM=one,bctc,dumla,countfo,yrsr,indv_id,group_id,no_ot_id;
Pts=3;Parameters;Pds=4;Matrix;Crosstab;Prob=prLC2a$

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$Summary Statistics (mean) Used for LCM Willingness-to-Accept Estimates
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

DSTAT;Rhs=bctc,dumla,countfo,yrsr,indv_id,group_id,no_ot_id,erscb,ersnp,ersnwa,
erssea,erssp$

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

197

REFERENCES

198

REFERENCES

Adamowicz, W., R. Boxall, M. Williams and J. Louviere. “Stated Preference Approaches
for Measuring Passive Use Values: Choice Experiments and Contingent Valuation.”
American Journal of A gricultural Economics 80(1998):64-75.

Adamowicz, W., J. Louviere, J. Swait. “Introduction to Attribute-Based Stated Choice
Methods.” Submitted to Resource Valuation Branch Damage Assessment Center,
National Oceanic and Atmosphere Administration, and U.S. Department of
Commerce. Edmonton, Alberta, Canada. January 1998.

Alfnes, F. “Stated Preferences for Imported and Hormone-Treated Beef: Application of a
Mixed Logit Model.” European Review of Agricultural Economics 31(2004):19-37.

Ammendrup. S. and L0. Barcos. The implementation of traceability systems. Review of
Science Technology Off. int. Epiz., 25.2(2006):763-773.

Bailey, D. and J. Slade. “Factors Influencing Support for a National Animal Identification
System for Cattle in the United States.” Selected paper presented at the American
Agricultural Economics Association Annual Meetings. Denver, CO. August 1-4,
2004.

Baines, R. ‘N., and W. P. Davies. “Quality Assurance in the International Food Supply.”
Paper presented at the Third International Conference on Chain Management in
Agribusiness and the Food Industry, Wageningen University, Wageningen, The
Netherlands, 1998.

Bech, M. and D. Gyrd-Hansen. “Effects Coding in Discrete Choice Experiments.” Health
Economics 14(2005): 1079-1083.

Becker, GS. 2007. Animal Identification and Meat Traceability. CRS report prepared for
members and committees of congress. Retrieved from
http://www.nationalaglawcenter.org/assets/crs/RL32012.pdf on February 5, 2008.

Becker, T. “Consumer Perception of Fresh Meat Quality: A Framework for Analysis.”
British Food Journal 102.3(2000)2158-176.

Blumenschein, K., M. Johannesson, G.C. Blomquist, B. Liljas, R.M. O’Connor.
“Experimental Results on Expressed Certainty and Hypothetical Bias in Contingent
Valuation.” Southern Economic Journal 65.1(1998): 1 69-177.

Bockstael, N. E., and I. E. Strand, Jr. "The Effects of Common Sources of Regression
Error on Benefit Estimates." Land Economics 63(Feb 1987):]1-20.

199

Boxall, P.C., W.L. Adamowicz. “Understanding Heterogeneous Preferences in Random
Utility Models: A Latent Class Approach.” Environmental and Resource Economics
23(2002):421-446.

Brester, Gary W. “Meeting Consumer Demands with Genetics and Market Coordination:
The Case of the Leachman Cattle Company.” Review of Agricultural Economics 24.1
(2002):251-265.

Buhr, B. “Traceability and Information Technology in the Meat Supply Chain:
Implications for Firm Organization and Market Structure.” Journal of Food
Distribution Research 34.3(2003): 13-26.

Bulut, H., and J. D. Lawrence. 2007. “Meat Slaughter and Processing Plants’Traceability
Levels Evidence From Iowa.” Proceedings of the NCCC-134 Conference on Applied
Commodity Price Analysis, Forecasting, and Market Risk Management. Chicago, IL.
Retrieved from http://www.farmdoc.uiuc.edu/nccc134 on February 5, 2008.

Bunch, D. S. and R. Kitamura. “Probit Model Estimation Revisited: Trinomial Models of
Household Car Ownership.” Working Paper. University of California Transportation
Center. University of California Berkeley, CA. 1991.

Buzby, J. C., P. D. Frenzen, and B. Rasco. 2001. “Product Liability and Microbial
F oodboume Illness.” Economic Research Service, AER 799. Retrieved from
www.crs.usda.gov/pwblications/aer799/ on February 5, 2008.

 

Caswell, J .A. "Labeling Policy for GMOs: To Each His Own?" AgBioForum 9(2000):53-
57.

Cattle Network. 2008. “Five Minutes With Dr. John Wiemers, USDA, APHIS, NAIS.”
Retrieved from http://www.cattlenetwork.com/content.asp?ContentId=194952 on
February 5, 2008.

Champ, P., R.C. Bishop. “Donation Payment Mechanisms and Contingent Valuation: An
Empirical Study of Hypothetical Bias.” Environmental and Resource Economics.
19(2001): 383-402.

Dickinson, D and D. Bailey. “Meat Traceability: Are U.S. Consumers Willing to Pay for
It?” Journal of A griculture and Resource Economics 27.2(2002):348-364.

Dickinson, D. L., J. E. Hobbs, and D. Bailey. 2003. “A Comparison of U.S. and Canadian
Consumers’ Willingness To Pay for Red-Meat Traceability.” Paper prepared for

presentation at the American Agricultural Economics Association Annual Meetings.
Montreal, Canada. July 27-30, 2003.

Early, R. “Farm Assurance-Benefit or Burden?” Journal of the Royal Agricultural
Society 159(1998):32-43.

200

F ishbein, Martin, and lock Ajzen. Belief, Attitude, Intention, and Beh_avior: An
Introduction to Theory and Research. Addison-Wesley, 1975.

Golan, E., B. Krissoff, F. Kuchler, L. Calvin, K. Nelson, and G. Price. 2004.
“Traceability in the U.S. Food Supply: Economic Theory and Industry Studies.”
Agricultural Economic Report Number 830, Economic Research Service, United
States Department of Agriculture. Retrieved from
htthwwersusdagov/Publications/AER830/ on February 5, 2008.

Golan, E., B. Krissoff, F. Kuchler, K. Nelson, G. Price and L. Calvin. “Traceability for
Food Safety and Quality Assurance: Mandatory Systems Miss the Mark. Current
Agriculture, Food and Resource Issues 4(2003):27-35.

Gong, W., k. Parton, R. J. Cox, and Z. Zhou. “Transaction Costs and Cattle Farmers’
Choice of Marketing Channels in China, A Tobit Analysis.” Management Research
News 30.1(2007):47-56.

Greene, W.H. Econometric Analysis. 4th ed. New Jersey: Prentice-Hall Inc, 2000.

 

Gregory, A. 2008. “Optimal Economic Design of Mail Surveys: Influences of Response
Rates and the Impact of Responses to a Second Mailing.” A Disseration. Kansas State
University, Department of Agricultural Economics.

Gujarati, Damodar N. Basis Econometrics. 4th ed. New York: McGraw-Hill/Irwin, 2003.

 

Hall, J ., Viney, R., Haas, M., & Louviere, J. 2004, "Using stated preference discrete
choice modeling to evaluate health care programs", Journal of Business Research
57.9(2004): 1026-1032.

Han, J. and R. W. Harrison. 2006. “Consumer Valuation of the Second Generation of
Genetically Modified (GM) Foods with Benefits Disclosure.” Selected Paper
prepared for presentation at the Southern Agricultural Economics Association Annual
Meetings. Orlando, Florida. February 4-8, 2006.

Hardy, M. A., A. Bryman. Handbook of Data Analysis. Sage Publications Inc, 2004.

Hensher, D. A., J. M. Rose, and W. H. Greene. Applied Choice Analysis A Primer.
Cambridge, UK: Cambridge University Press, 2005

Hobbs, J .E. “Traceability in Meat Supply Chains, Current Agriculture, Food and
Resource Issues” 4(2003a):36-49.

Hobbs, Jill E. 2003b. “Consumer Demand for Traceability.” Working Paper #03-1,

International Trade Research Consortium (IATRC). Retrieved from
http://www.iatrcwebbrg on February 5, 2008.

201

 

Hobbs, J. E. 2003c. “Traceability and Country of Origin Labeling.” Farm Policy
Development and Tensions with NAFTA. University of Saskatchewan.
Saskatchewan, Canada.

Hobbs, J .E. “Information Asymmetry and the Role of Traceability Systems.”
Agribusiness 20.4(2004):397-415.

Hobbs, J .E. “Liability and Traceability in Agri-Food Supply Chains.” In Ondersteijn,
CJ M, Wijnand, J HM, Huirne, RBM and van Kooten, O, editors. Quantifying the
A gri-F ood Supply Chain. Netherlands: Springer, (2006):85-100.

Hobbs, J .E, D. Bailey, D.L. Dickinson, and M. Haghiri. “Traceability in the Canadian
Red Meat Sector: Do Consumers Care?” Canadian Journal of Agricultural
Economics; 53.1(2005):47-65.

Huffman, WE., M. Rousu, J .F. Shogren, and A. Tegene. "Consumer Willingness to Pay
for Genetically Modified Food Labels in a Market with Diverse Information:
Evidence from Experimental Auctions." Journal of A gricultural and Resource
Economics 28.3(2003a):481-502.

Huffman, WE., M. Rousu, J.F. Shogren, and A. Tegene. "The Public Good Value of
Information from Agribusinesses on Genetically Modified Foods." American Journal
of A gricultural Economics 85.5(2003b): 1309-13 15.

International Organization for Standardation. 2000. “Quality management systems-
Requirements.” ISO 9001. Retrieved from
http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detai1.htm?csnumber=2
1823 on July 18, 2008.

Jensen, H. H. and D. J. Hayes. 2006. “Private Sector Approaches to Secure Traceability,
Transparency and Quality Assurance in Food Chains.” Paper prepared for the IATRC
Summer symposium "Food Regulation and Trade: Institutional Framework, Concepts
of Analysis and Empirical Evidence". Bonn, Germany, May 28-30, 2006.

Kansas State University. 2006. “BEEF and K-State Cow-Calf Producer NAIS Survey.”
By Prism Media. Department of Animal Science. Manhattan, Kansas.

Kennedy, P. A Guide to Econometrics. 5th ed. Cambridge, MA: The MIT Press, 2003.

Krinsky, I., and A. L. Robb. "On Approximating the Statistical Properties of Elasticities."
Review of Economics and Statistics 68 (Nov 1986):715-19.

Kuhfeld, W.F., R.D. Tobias, and M. Garratt. “Efficient Experimental Design with

Marketing Research Applications.” Journal of Marketing Research 31(1994):545-
557.

202

Lancaster, K. 1966. “New Approach to Consumer Theory.” Journal of Political Economy
74(1966):132-57.

Layton, DE 1996. “Rank-Ordered, Random Coefficients Multinomial Probit Models for
Stated Preference Surveys.” Paper Presented at the Association of Environmental and
Resource Economists Workshop. Tahoe City, California. June 2-4, 1996.

Liddell, S. and D. Bailey. 2001. “Market opportunities and threats to the U.S. pork
industry posed by traceability systems.” International Food and Agribusiness,
4(2001): 287-302.

Loureiro, M., and W. Umberger. “Estimating Consumer Willingness to Pay for Country-
of-Origin Labeling.” Journal of Agriculture and Resource Economics 28(2003):287-
301.

Lupin, H. 2006. “Traceability.” Presented for Discussion at the Committee On Fisheries,
and Sub-Committee On Fish Trade (COFleT). Tenth Session. Santiago de
Compostela, Spain, May 30-June 02, 2006.

Lusk, J .L. “Willingness-to-Pay for Golden Rice.” American Journal of A gricultural
Economics 85.4(2003):840-856.

Lusk, J .L., F .B. Norwood, J .R. Pruitt. “Consumer Demand for a Ban on Antibiotic Drug
Use in Pork Production.” American Journal of Agricultural Economics.
88.4(2006):1015-1033.

Lusk, J ., J. Roosen, and J. Fox. “Demand for Beef from Cattle Administered Growth
Hormones or Fed Genetically Modified Corn: A comparison of Consumers in France,
Germany, the United Kingdom, and the United States.” American Journal of
Agricultural Economics 85(2003): 16-29.

McFadden D. “Econometric models of probabilistic choice.” In C.F. Manski and D.
McFadden, editors. Structural Analysis of Discrete Data with Econometric
Applications. pp. 198-272. Cambridge, MA: The MIT Press, 1981.

Mennecke B., A. Townsend, D.J. Hayes, and S. Lonergan. 2006. “A Study of the Factors
that Influence Consumer Attitudes Toward Beef Products Using the Conjoint Market
Analysis Tool.” Iowa State University, CARD Research Paper 06-WP 425. August,
2006.

Meuwissen, M. P. M., A. G. J. Velthuis, H. Hogeveen, and R. B. M. Huirne. 2003.

“Traceability and Certification in Meat Supply Chains.” Journal of Agribusiness
21.2(2003):167-181.

203

 

Millimet, D. 2001. “Endogeneity versus sample selection bias. STATA, Data Analysis
and Statistical Software.” Retrieved from
http://www.stata.com/sugporflfaQ§/stadbias.html on February 5, 2008.

Morey, E.R. “TWO RUMS un CLOAKED: Nested-Logit Models on Site Choice and
Nested-Logit Models of Participation and Site Choice.” In Herriges, J .A. and CL.
Kling editors. Valuing Recreation and the Environment. Northampton, MA: Edward
Elgar, 1999.

Mus, M. “Traceability System Approaches and Cost Analysis for the Beef Industry.” A
Thesis. Washington State University, School of Economics Sciences. 2006

Nahuelhual, L., M.L. Loureiro, and J. Loomis “Using Random Parameters to Account for
Heterogeneous Preferences in Contingent Valuation of Public Open Space.” Journal
of Agricultural and Resource Economics. 29(2004):537-5 52.

Nakosteen, R., and M. Zimmer. “Migration and Income: The question of Self-Selection.”
Southern Economic Journal 46(1980):840-851 .

National Agriculture Statistics Service (NASS). 2008. United States Department of
Agriculture, Quick Stats. Retrieved from
http://www.nass.usda.gov/OuickStats/Create_Federal All.isp. on January 23, 2008

New Zealand Trade and Enterprise. 2006. “Traceability a Platform for Product
Knowledge and Value.” A Report for New Zealand Trade and Enterprise. June, 2006.

Ouma, E., A. Abdulai, and A. Drucker. “Measuring Heterogeneous Preferences for Cattle
Traits among Cattle-Keeping Households in East Africa.” American Journal of
Agricultural Economics 89.4(2007): 1005-1019.

Park, T., J. Loomis, and M. Creel. "Confidence Intervals for Evaluating Benefits From
Dichotomous Choice Contingent Valuation Studies." Land Economics 67.1(1991):64-
73.

Pindyck, RS. and D.L. Rubinfeld. Econometric Models and Economic Forecasts. 4th ed.
Boston, Massachusetts: The McGraw-Hill Companies Inc, 1998.

Revelt, D., and K. Train. “Mixed Logit with Repeated Choices: Households’ Choices of
Appliance Efficiency Level.” Review of Economics and Statistics 80(1998):647-657.

Rigby, D. and M. Burton. “Preference Heterogeneity and GM Food in the UK. European
Review of A gricultural Economics. 32(2005):269-288.

Roberts, T., Busby, J .C., and M. Ollinger. “Using Benefit and Cost Information to

Evaluate a Food Safety Regulation: HACCP for Meat and Poultry”, American
Journal of A gricultural Economics 78(1996): 1297-1301 .

204

Sanderson, Kim and J. E. Hobbs. 2006. “Traceability and Process Verification in the
Canadian Beef Industry.” Report prepared for Canfax Research Services. October,
2006.

SAS Institute Inc., SAS 9.1.3 Help and Doc. Cary, NC: SAS Institute Inc., 2008.

Scarpa, R. and T. Delgiudice. Market Segmentation via Mixed Logit: Extra Virgin Olive
Oil in Urban Italy. Journal of Agricultural and Food Industrial Organization. Vol.2,
Article 7(2004).

Schroeder, T., G. Tonsor, J. Mintert, J .M.E. Pennings. “Consumer Risk Perceptions and
Attitudes about Beef Food Safety: Implications for Improving Supply Chain
Management.” Kansas State University Agricultural Experiment Station and
Cooperative Extension Service. 2006.

Schroeder et a1. 2007. “Benefit Cost Analysis of the National Animal Identification
System.” U.S. Department of Agriculture, Animal and Plant Health Inspection
Service. 2007

Small, K.A., and HS. Rosen. “Applied Welfare Economics with Discrete Choice
Models.” Econometricia 49(1978):43-6.

Smith, G.C., Belk, K.E., Scanga, J .A., Sofos, J .N., Tatum, JD. 2000. “Traceaback,
traceability and source verification in the US beef industry.” Presented at the XXI
World Buiatrics Congress, Punta del Este, Uruguay. December 5, 2000. Retrieved
From http://ansci.colostate.edu/dp_/msfs/gcsOO1 .pdf on January 23, 2008.

Smith, G.C., J .D. Tatum, K.E. Belk, J .A. Scanga, T. Grandin, and J .N. Sofos.
“Traceability from a US Perspective.” Meat Science 71(2005):174-193

Smyth, S., and P.W.B. Phillips. “Product Differentiation Alternatives: Identity
Preservation, Segregation, and Traceability.” AgBioForum 5.2(2002):30-42

Souza-Monteiro, D.M. and J .A. Caswell. 2004. The economics of implementing
traceability in beef supply chains: Trends in major producing and trading countries.
Presented at the Annual meeting of the Northeastern Agricultural and Resource
Economics Association. Halifax, Nova Scotia. June 20-23, 2004.

Starbird, S. A. and V. Amanor-Boadu. “Contract Sensitivity, Food Safety, and
Traceability.” Journal of Agriculture & Food Industrial Organization 5.2(2007).

The Food Business Forum. “Implementing Traceability in the Food Supply Chain.” pp. 5.
Paris, France. 2005.

Tobin, J arnes. “Estimation of Relationship for Limited Dependent Variables.”
Econometrica, 26(195 8):24-36.

205

Tonsor, GT. and TC. Schroeder. “Livestock Identification: Lessons for the U.S. Beef
Industry from the Australian System.” Journal of International Food & Agribusiness
Marketing, 18.3/4(2006):103-1 18.

Tonsor, G. T., T. C. Schroeder, J. A. Fox, and A. Biere. “European Preferences for Beef
Steak Attributes.” Journal of Agricultural and Resource Economics 30.2(2005):367-
380.

Tonsor, G.T., T.C. Schroeder, J .M.E. Pennings, and J. Mintert. 2008. “Consumer
Valuations and Choice Processes of Food Safety Enhancement Attributes: An
International Study of Beef Consumers.” Working paper.

Train, K.E. “Mixed Logit Models for Recreation Demand.” in C. Kling and J. Herriges,
editors, Valuing the Environment Using Recreation Demand Models. Lyme, New
Hampshire: Edward Elgave, 1997.

Train, K.E. “Recreation Demand Models with Taste Differences over People.” Land
Economics 74(1998):230-239.

 

Train, K. E., Discrete Choice Methods with Simulation. Cambridge, MA: Cambridge
University Press, 2003.

United States Department of Agriculture, Animal and Plant Health Inspection Service.
2008. “National Animal Identification System.” Retrieved from
http://animalid.aphis.usda.gov/nais/index.shtml on July 8, 2008.

United States Government Accountability Office. “National Animal Identification,
USDA Needs to Resolve Several Key Implementation Issues to Achieve Rapid and
Effective Disease Traceback.” Report GAO-07-592 to the Chairman, Committee on
Agriculture, Nutrition, and Forestry, U.S. Senate. 2007.

Weiss, M. D. 1995. “Information Issues for Principals and Agents in the Market for
Food Safety and Nutrition.” pp 69-79 in J. A. Caswell, editors, Valuing Food Safety
andNutrition. Boulder, CO: Westview Press, 1995.

Wooldridge, Jeffrey M. Introductory Econometrics, A Modern Approach. 2nd ed. Mason,
Ohio: South-Western College Publishing, 2003.

Wooldridge, Jeffrey M. Introductory Econometrics, A Modern Approach. 3rd E. Mason,
Ohio: South-Western College Publishing, 2006.

206

 

 

111111111111131131111I