© 2014 xiang cao...xiang cao august 2014 chair: lisa house cochair: zhengfei guan major: food and...
TRANSCRIPT
1
ADDRESSING PRODUCTION AND MARKETING CHALLENGES OF THE FLORIDA TOMATO INDUSTRY
By
XIANG CAO
A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
2014
2
© 2014 Xiang Cao
3
ACKNOWLEDGMENTS
I would like to thank the Food and Resource Economics Department (FRED) for
offering me the chance to study in the United States. Without the professional training
provided by FRED, this research would have been difficult.
Foremost, I would like to express my sincere gratitude to my committee chair
Dr. Lisa House and co-chair Dr. Zhengfei Guan for their continuous support of my
master’s study and research, and for their patience, motivation, enthusiasm, and
immense knowledge. Their guidance helped me throughout the research and writing of
this thesis. It would have been much harder to achieve my degree without their advice
and guidance.
Other thanks also go to my committee members, Dr. Zhifeng Gao and Dr. Gary
Vallad, for their encouragement, instructions, and insightful comments. Their
tremendous help ensured the successful completion of this thesis.
In addition, I would also like to thank other faculty, staff, and my classmates in
FRED. Their help and support made my life at UF enjoyable and meaningful.
I also want to thank Dr. Feng Wu who offered me invaluable help in my research.
His kindness and wisdom enlightened me and enriched my academic experience.
Last but not the least, I would like to thank my parents, for giving me life, raising
me, and nurturing me spiritually. And I also would like to thank my girlfriend Yunjia, for
her kindness, companionship, and support all the time.
4
TABLE OF CONTENTS page
ACKNOWLEDGMENTS .................................................................................................. 3
LIST OF TABLES ............................................................................................................ 6
LIST OF FIGURES .......................................................................................................... 7
LIST OF ABBREVIATIONS ............................................................................................. 8
ABSTRACT ..................................................................................................................... 9
CHAPTER
1 INTRODUCTION .................................................................................................... 11
Background Information .......................................................................................... 11
Objectives ............................................................................................................... 14 Organization ........................................................................................................... 15
2 LITERATURE REVIEW .......................................................................................... 18
3 ECONOMIC ANALYSIS OF FUMIGATION ALTERNATIVES, THE METHYL BROMIDE BAN, AND ITS IMPLICATION ............................................................... 28
Fumigation Information in the Florida Tomato Industry ........................................... 28 Materials and Methods............................................................................................ 30
Partial Budgeting Analysis ................................................................................ 31 Stochastic Dominance and Stochastic Efficiency Analysis ............................... 31
Data Description and Adjustment ........................................................................... 36
Data Description ............................................................................................... 36 Data Adjustment ............................................................................................... 39
Results .................................................................................................................... 40 Fumigation Costs of MBr and Alternative Soil Treatments ............................... 40 Estimated Yields, Total Costs, and Net Returns Associated with MBr and Its
Alternative Treatments .................................................................................. 41
Stochastic Dominance and Stochastic Efficiency Analysis Results .................. 44
4 DETERMINING THE IMPACT OF STATE-SPECIFIC SIGNS AND LABELS ON TOMATO MARKETING .......................................................................................... 54
Survey Designing and Data Collection.................................................................... 54 Descriptive Analysis of the Survey Data ................................................................. 59
Demographic of Participants ............................................................................ 59 Consumers’ Purchasing Habits of Fresh Tomatoes ......................................... 60 Consumers’ Attitudes and Preference of Different Labeled Fresh Tomatoes ... 61
5
Consumers’ WTP for Florida/US Tomatoes and Mexico Tomatoes ................. 62 Models .................................................................................................................... 62 Results .................................................................................................................... 66
Factors Influencing Whether or Not Consumers Read Tomato Cool ................ 66 Factors Affecting Consumer’s Choice for Tomatoes with Different Labels ....... 68 Factors Affecting Consumer’s WTP: A Premium for Florida/US Tomatoes ...... 70
5 CONCLUSIONS ..................................................................................................... 77
LIST OF REFERENCES ............................................................................................... 81
BIOGRAPHICAL SKETCH ............................................................................................ 89
6
LIST OF TABLES
Table page 3-1 Regression results on tomato yield adjustments from SAS ................................ 46
3-2 Estimated average tomato fumigation costs for MBr:Pic (67:33) and selected alternative soil treatments and the fumigation costs of the alternative treatments relative to MBr:Pic (67:33) ................................................................ 46
3-3 Marketable tomato yields, gross revenue for MBr:Pic (67:33) and selected alternative fumigant treatments, and the difference in gross revenues relative to MBr:Pic (67:33) ............................................................................................... 47
3-4 Estimated average costs incurred to produce, fumigate, and harvest tomatoes, total net returns, and net returns of the selected alternative soil treatments relative to MBr:Pic (67:33) ................................................................ 47
3-5 Total negative effects (added costs and reduced returns), total positive effects (reduced costs and added returns), and total effects of the selected alternative soil treatments relative to MBr:Pic (67:33) in the tomato production system .............................................................................................. 48
3-6 Second-order stochastic dominance result of adjusted yield of methyl bromide and its alternative treatments ................................................................ 49
3-7 Summary statistics of net returns of MBr:Pic (67:33) and its alternatives ........... 49
4-1 Sample demographic descriptive statistics ......................................................... 72
4-2 Consumers’ stated choice of different labeled tomatoes, sorted by scenario and by city .......................................................................................................... 73
4-3 Willingness to pay for Florida/US and Mexico tomatoes, sorted by scenario and by city (Unit: $/lb) ......................................................................................... 73
4-4 Parameter results of the binary logistic regression of consumers’ behavior of reading tomato COOL information in the experiment ......................................... 74
4-5 Parameter results of the ordered logistic regression of consumers’ stated choice of tomatoes from the first simultaneous equation system........................ 75
4-6 Parameter results of the linear regression of consumers’ willingness to pay for a premium for Florida/US tomatoes than for Mexico tomatoes from the second simultaneous equation system ............................................................... 76
7
LIST OF FIGURES
Figure page 1-1 US and Florida fresh tomatoes productions from 2000 to 2012 ......................... 16
1-2 Origins of US fresh tomatoes imports (Mexico, Canada, and other countries) ... 16
1-3 Florida annual pre-plant methyl bromide usage info before year 2000 ............... 17
3-1 Explanatory figure of expected money value, certainty equivalent, and risk premium ............................................................................................................. 50
3-2 Comparison of MBr:Pic (67:33) and its alternatives’ CDF series for adjusted yield data ............................................................................................................ 51
3-3 Comparison of the SERF results of the six treatments’ net returns under power utility function ........................................................................................... 52
3-4 Comparison of the six fumigation treatments’ risk premiums under the power utility function relative to the non-fumigated treatment ($/acre) .......................... 53
8
LIST OF ABBREVIATIONS
CE
COOL
CVM
MBr
RP
Certainty Equivalent
Country of Origin Labeling
Contingent Valuation Method
Methyl Bromide
Risk Premium
SD Stochastic Dominance
SSD
SERF
WTP
Second-Order Stochastic Dominance
Stochastic Efficiency with Respect to a Function
Willingness to Pay
9
Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science
ADDRESSING PRODUCTION AND MARKETING CHALLENGES
OF THE FLORIDA TOMATO INDUSTRY
By
Xiang Cao
August 2014
Chair: Lisa House Cochair: Zhengfei Guan Major: Food and Resource Economics
The technological shock from the methyl bromide phase-out and the intense
competition from Mexico have posed serious threats and challenges to the Florida
tomato industry in both production and marketing. This research contains two main
parts. The first part focuses on identifying an optimal fumigation strategy through
analyzing the cost effectiveness and risk efficiency of methyl bromide treatment and its
alternatives. Partial budgeting and stochastic dominance analyses are performed based
on the data acquired from scientific field trials. The second part provides the struggling
US tomato industry with marketing information to understand consumer demand and
willingness to pay for local (Florida/US) tomatoes versus non-local (Mexico) tomatoes.
In addition, the effect of three market strategies on consumer preferences for Florida/US
versus Mexico tomatoes is also studied. A mall intercept survey using the contingent
valuation method to interview 632 participants was conducted to determine US
consumer perception about country of origin labeling, consumption pattern, demand,
and willingness to pay for Florida/US and Mexico tomatoes. This thesis research
10
provides recommendations to help the industry address challenges from both
production and marketing perspectives.
11
CHAPTER 1 INTRODUCTION
Background Information
The United States (US) has been one of the world’s largest tomato producers for
decades. According to the data from the Food and Agricultural Organization (FAO) of
the United Nations, the United States was the second largest tomato producer (behind
China) before 2011, and the third largest tomato producer (behind China and India) in
2011 and 2012 (UN/FAO, 2013). Tomato production generates the highest value among
all the vegetable crops grown in the United States (USDA/ERS, 2013). In 2012, the
United States produced 27.6 million hundred weight (cwt) fresh tomatoes, valued at
$0.86 billion (USDA/NASS, 2013a). The top three fresh-tomato-producing US states are
Florida, California, and North Carolina. Florida has been the largest US fresh tomato
supplier for decades. In 2012, Florida produced nearly 9.6 million cwt fresh tomatoes,
worth $268 million, accounting for 34.8% and 31.2% of total US production,
respectively. Fresh tomato yield in Florida ranks the highest among all US states,
averaging roughly 330 cwt per acre (USDA/NASS, 2012a).
Over the past decade, however, US fresh tomato production has been declining,
from 40 million cwt in 2002 to 28 million cwt in 2012. Florida production has also
decreased, from 14 million cwt to 9.6 million cwt (Figure 1-1) (USDA/NASS, 2013a).
The value of US fresh tomato production dropped from $1.4 billion to $0.86 billion over
the same period. In Florida, harvested acreage has decreased as well, from the
historical high of 45 thousand acres in 2001 to 29 thousand acres in 2012.
One of the main reasons identified for the decreases in the US and Florida
tomato industries is competition from Mexico. Mexico’s competitive advantage in tomato
12
production cost and favorable government policies has encouraged fresh tomato
exports to the US domestic market. Specifically, fresh tomato production cost is much
lower in Mexico than in the United States due to lower labor costs. In addition, Mexico
has encouraged investment in agriculture to upgrade its technology to improve its
production capacity, such as its protected greenhouse tomato production.
Historically, Florida and Mexico have competed for the US winter and early
spring tomato market. Tomato imports from Mexico reach their peak in winter when
Florida is the predominant domestic tomato producer. Florida harvests fresh tomatoes
from October to June each year, reaching peak production from November to January.
In winter, most Florida tomatoes are shipped to the eastern United States while Mexico
fresh tomatoes are shipped to the western United States (VanSickle et al., 2003).
Under the North American Free Trade Agreement (NAFTA) enacted in 1994,
Mexico has increased its fresh tomato exports to the United States, from 13 million cwt
in 2000 to 30.4 million cwt in 2012, based on data from the United States Department of
Commerce (Figure 1-2) (USDOC, 2013). The data indicate that imports from Mexico
now account for about 90% of the imported tomatoes to the United States and have had
a major impact on the US tomato industry, particularly the Florida tomato industry.
Mexico tomatoes sold on the market were about 20% less than Florida’s supply volume
in 2000, but their market share is now more than three times higher than Florida’s
market share (Zhu et al., 2013). As competition from Mexico increased, the farm gate
value of the Florida tomato industry slumped from $620 million in 2010 to $268 million in
2012, the lowest value during the last decade (USDA/NASS, 2013a).
13
As one of the most important agricultural industries in the state, the Florida
tomato industry also is facing increasing challenges due to the costs of domestic
production. Tomato production costs in Florida continue to increase, with a main
contributor being soil fumigation costs. Soil fumigation is applied to control nematodes,
soil-borne pests, weeds, and plant diseases. Methyl bromide (MBr), which has now
been banned in the United States, has been shown to be the most effective soil
fumigant. In the past, most US agricultural producers utilized MBr as the primary
fumigation strategy due to its easy operation and high efficacy. As shown in Figure 1-3,
many crops in Florida, including fresh tomatoes, peppers, and strawberries, accounted
for 50%, 32%, and 12% of total MBr pre-plant usage, respectively, in the United States
before 2000 (Osteen, 2000).
After MBr was identified as one of the chemical substances that can cause ozone
depletion, it was slated for phase-out in the United States under the Montreal Protocol
of 1987. In 1992, MBr was officially listed as a stratospheric ozone-depleting substance
and, in 1997, the Montreal Protocol required that developed countries must eliminate all
MBr fumigants by 2010, with few exceptions. In the United States, MBr is now only
permitted under Critical Use Exemptions (CUEs) on a very limited scale under close
government scrutiny. Between 2005 and 2013, CUEs were authorized for Florida
tomato, strawberries, pepper, and eggplant commercial production.
Since the MBr phase-out, US growers have gradually started applying other
fumigants in place of MBr, but scientific research has been unable to find feasible
fumigant alternatives with the same consistent, high technical effectiveness and low
cost as MBr. Alternative fumigants currently used in the US agricultural industry have
14
exhibited significant variations under different conditions. The MBr phase-out has put
tremendous pressure on the US fruit and vegetable industries. The Florida tomato
industry, the largest fresh tomato supplier in the nation, has suffered from increasing
production costs and decreasing yields. Economically, the MBr phase-out has caused
substantial financial losses to Florida producers, with estimated losses calculated at
more than $500 million (Spreen et al., 1995). According to the fumigation usage survey
conducted in 2012 by the Florida Tomato Exchange and the Florida Fruit and Vegetable
Association (FFVA), the new MBr alternatives/substitutes have caused up to a 20%
yield loss compared to MBr treatments, and growers have suffered from extra costs
resulting from additional Integrated Pest Management (IPM) under the current
fumigation system.
Between 2005 and 2013, CUEs were authorized in Florida for commercial tomato
production, but CUEs expired at the end of 2013. A petition submitted for an exemption
to the rule for Florida in 2013 was rejected. The petition had requested additional MBr
allocations for 2015 and 2016 as a “methyl bromide rescue treatment” (as formulated
with chloropicrin MBr:Pic (67:33)) in critical situations where the chemical alternatives
failed to manage soilborne pests, pathogens, and weeds.
Objectives
The technological shock from the MBr phase-out in the United States, coupled
with intense competition from Mexico (classified as a developing country still allowed to
use methyl bromide under the Montreal Protocol) has challenged and imposed
pressures on the Florida tomato industry. Against such a background, the first objective
of this thesis is to provide growers and policy makers with scientific information on
optimal fumigation strategies, so as to help lower production costs and risks and to
15
boost yields in the struggling Florida tomato industry. We will evaluate the cost
effectiveness and risk efficiency of MBr and MBr alternatives for the Florida tomato
industry using partial budgeting analysis and stochastic dominance methods based on
field trials conducted by the University of Florida.
A second objective of this thesis is to provide the struggling US tomato industry
with marketing information to understand consumer demand and willingness to pay for
local (Florida/US) tomatoes, as consumer choice is of vital importance to the domestic
tomato industry. Based on a mall intercept survey using the Contingent Valuation
Method (CVM) conducted by the University of Florida, we will study the effect of three
market strategies on consumer preference for Florida/US tomatoes versus Mexico
tomatoes. Country of origin labeling effect, consumer preference, and willingness to pay
will be identified and estimated through ordinal logistic and simple linear regressions.
Organization
The remainder of this thesis is divided into four chapters. Chapter 2 is the
literature review with respect to the economic impact of the MBr phase-out based on
partial budgeting analysis, stochastic dominance studies, country of origin labeling
(COOL) research, and “local food” studies. Chapter 3 presents the results of the cost
effectiveness and risk efficiency of MBr alternatives in Florida tomato production
through partial budgeting and stochastic dominance analyses, including experiment
data description, theoretical statements, and results analysis. Chapter 4 mainly focuses
on determining the impact of state-specific signs and labels on tomato marketing,
including data description, CVM statements, regression modeling, and results analysis.
Chapter 5 summarizes the studies, presents the conclusions and discussion, addresses
limitations, and offers suggestions for future research.
16
Figure 1-1. US and Florida fresh tomatoes productions from 2000 to 2012
Source: USDA/NASS (2013a).
Figure 1-2. Origins of US fresh tomatoes imports (Mexico, Canada, and other
countries)
Source: USDOC (2013).
38.8937.70
39.59
35.3637.95 38.03
36.2733.63
31.1433.24
27.96 28.23 27.59
15.76 14.91 13.98 14.19 15.12 15.5413.48 13.32
10.4612.30
8.56 9.12 9.57
0
5
10
15
20
25
30
35
40
45
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Mill
ion c
wt US total Florida
13.0014.97 15.95 17.31 17.17 17.67 18.61
20.94 21.77 23.08
30.43 29.26 30.43
0
5
10
15
20
25
30
35
40
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Mill
ion c
wt
Mexico Canada Others
17
Figure 1-3. Florida annual pre-plant methyl bromide usage info before year 2000
Source: USDA/ERS (2013a).
50%
32%
12%
6%
Tomatoes Peppers Strawberries Other
18
CHAPTER 2 LITERATURE REVIEW
There have been some prior studies on the economic impact, feasibility, and
efficacy of MBr and its fumigant alternatives after the MBr phase-out. The majority of
these studies focus on estimating the economic impact that resulted from the MBr ban.
Other research discussed the economic viability other than the technical feasibility of
various MBr substitutes/alternatives to find out whether there exist optimal fumigation
strategies to replace MBr.
In 1993, the National Pesticide Impact Assessment Program (NAPIAP) of the
United States Department of Agriculture estimated annual economic losses of $1.3
billion to $1.5 billion if an MBr ban occurred in the United States, of which $900 million
would be attributed to soil fumigation. The NAPIAP also estimated that the greatest
losses would occur in tomatoes ($350 million) and strawberries ($110 million). Specific
estimated impacts to the state of Florida included a 45% to 50% decrease in tomato
production and a 65% to 70% decrease in strawberry production (NAPIAP, 1993).
Deepak et al. (1996) estimated the influence of the MBr phase-out on the winter
market for fresh vegetables, particularly on the Florida market. Florida is the leading US
domestic fresh vegetables supplier in winter due to Florida’s unique climate. Deepak et
al. (1996) used a quadratic and programming model and empirical specifications to
divide Florida into four regions and to analyze the per acre production cost and yield
data with and without MBr. They then compared the per acre production costs and
yields for selected crops such as tomatoes, pepper, cucumbers, squash, and eggplant
in Mexico and Texas, Florida’s two main competitors in winter fresh vegetable market
(Deepak et al., 1996). Empirical results suggested that the MBr ban would have a
19
negative impact on Florida fruit and vegetable producers. Tomato production in Florida
would decrease sharply, and Mexico would gain much of the market lost to Florida
because producers in Mexico would not be affected by the MBr phase-out. Consumers
would face higher retail prices, ranging up to 10%, depending on the commodity, month,
and location of the market (Deepak et al., 1996).
Lynch et al. (1997) systematically analyzed the economic impact of the MBr ban.
They identified and evaluated the costs and efficacy of the technically feasible chemical
and non-chemical alternatives to MBr in California and Florida. They concluded that if
1,3-Dichloropropene (Telone®), Chloropicrin, and Pebulate were available, the
dependence of Florida tomato growers on one single chemical combination would be
decreased, and the economic losses to Florida growers would be much smaller than
previously calculated (Lynch et al., 1997). Hueth et al. (1997) estimated the short-term
impact of eliminating MBr completely on California agriculture. Their study analyzed the
cost of the MBr ban in California via measuring the direct change in consumer and
producer welfare. The results showed that the impact on growers’ profits might vary
corresponding to the exotic pest infestation (Hueth et al., 1997). They further
acknowledged that the result had some uncertainties due to lack of enough
experimental trials and the incapacity to predict future outcomes.
Lynch and Carpenter (1999) regarded the MBr phase-out as a spatial partial
equilibrium problem and used two methods to investigate its impacts. The first method
“calculated the value of each pound of methyl bromide based on anticipated yield and
cost changes assuming constant prices” (Lynch & Carpenter, 1998, p.1). The second
method “examined the annual crops that use MBr extensively, allowing for both acreage
20
and price adjustments” (Lynch & Carpenter, 1998, p.1). The result showed that
replacing MBr with alternative fumigants would mean significant financial losses for
producers and higher prices and lower quality produce for consumers.
Carpenter et al. (2000) developed a comprehensive report about the impact of
the MBr phase-out. The main purpose of this study was to estimate the impact of an
MBr ban for different crops in different regions, including Florida. The research analyzed
the cost effectiveness and yield performance per acre for various MBr alternative
fumigants through field trials and expert opinions. The economics model in the report
included calculations of baseline equilibrium production, monthly shipment information
between production areas and markets, and monthly consumption in each
representative market in each month given the current technology (Carpenter et al.,
2000). The estimated potential production losses for the Florida tomato industry
resulting from the MBr ban ranged from 20% to 40%. Telone C-17 was expected to
substitute for MBr and various herbicides at an increased cost of $227.50 per acre for
Florida tomato and strawberry producers. Carpenter et al. (2000) also stated that
Florida producers would increase planted strawberry acreage and decrease planted
tomato and pepper acreages. Total net loss in welfare for the United States was
estimated at $76.5 million on the basis of using revenue to measure the overall
influence to all US producers and consumers.
VanSickle et al. (2000) expanded the study based on the 1996 research, with a
special emphasis on evaluating the impact of the MBr phase-out on the US fresh
vegetable industry. According to the costs and yields from the model, they concluded
that the expected impact on the US market due to the 2005 elimination of MBr would
21
have the greatest negative impact on the US strawberry market, and that tomato and
pepper production also would experience large financial losses. It was estimated that
Florida tomato growers would lose $68.8 million in shipping point revenues and that the
expected fumigation strategy would be Telone C17 plus herbicide Tillam (VanSickle et
al., 2000).
Sydorovych et al. (2008) investigated the cost-effectiveness of MBr and its
substituted fumigant treatments used in tomato and strawberry production in North
Carolinavia partial budgeting analysis. They collected the input cost and yield data from
field experiments and expert knowledge. In their study, MBr was the base treatment to
be compared with other fumigation strategies, including Telone-C35, Telone II,
Chloropicrin, Midas, Inline, etc. (Sydorovych et al., 2008). The results indicated that
there were technically and economically feasible alternatives to MBr for tomato
production in growing conditions similar to Fletcher, North Carolina (Sydorovych et al.,
2008).
Byrd et al. (2007), Ferrer (2010), and Fonash et al. (2010) studied the effects of
MBr alternatives with Georgia peppers. Using stochastic dominance and multi-period
programing methods to identify the feasible fumigant-herbicide system alternative to
MBr for Georgia pepper producers, Byrd et al. (2007) found that a Telone II and
chloropicrin combination with metham potassium might offer a viable substitute for MBr.
Using the complete factorial treatment analysis approach, Ferrer et al. (2010) analyzed
the profitability efficiency of MBr and its alternatives for the Georgia pepper industry.
Their results indicated that the combination of 1,3-dichloropropene plus chloropicrin,
metham sodium, and smooth low density black on black polyethylene mulch was the
22
most profitable fumigant and mulch option. Similarly, Fonash et al. (2010) utilized the
multiple factor analysis to analyze comparative yield efficiencies of MBr alternatives and
mulching systems for pepper production in the southeastern United States. They
concluded that 1,3-dichloropropene plus chloropicrin plus metham sodium could
maximize pepper production.
In conclusion, previous research has proven that the MBr phase-out has resulted
in huge financial losses to the US fresh fruit and vegetable industries, including the
Florida tomato industry. Furthermore, different studies have identified various possible
MBr fumigation substitutes based on scientific experiments conducted in different US
locations.
The above review focused on fumigation and pest management on the
production side. On the marketing side, the review will focus on agricultural marketing
and consumer behavior. As an important food-product quality attribute, country of origin
labeling (COOL) has been studied for decades. This literature review starts with
previous research on food labeling, followed by prior COOL and local food studies and
the factors influencing consumer preference and willingness to pay for COOL.
Driven by increasing consumer demand for healthier, safer, and more
environmentally friendly food products, food labeling is playing an increasingly important
role in the food marketing system (McCluskey & Loureiro, 2003). Consumers acquire
various types of information about food-product attributes from food labels, thus helping
consumers make final purchasing decisions. Theoretically, consumers demand food-
product attributes (e.g., taste, nutrition, or food quality) rather than the food-product
itself. The food-product is simply regarded as a bundle of tangible or intangible
23
attributes that satisfies consumers and increases their utility. In 1966, Lancaster pointed
out that specific food-product attributes embodied in a food-product offer the basis for
purchasing decisions made by consumers. Thus food-product attributes via labels affect
consumer preference and willingness to pay (WTP) and, in turn, the demand for food
products.
Some food-product quality attributes are easily discernible by consumers,
especially for fresh produce, such as size, color, firmness, etc. These readily discernible
dimensions are called search and experience attributes (Stigler, 1961; Nelson, 1970).
Search attributes, referring to the visual attributes of the food-product (e.g., size, color),
can be obtained by consumers before purchasing. Experience attributes, such as taste,
are affirmed after consuming the food-product (Stigler, 1961; Nelson, 1970). However,
other quality features, such as organic, country of origin, and locally grown, cannot be
ascertained by consumers through direct experience. These attributes are defined as
credence attributes. Credence attributes cannot be evaluated by consumers before
purchase or after consumption without incurring prohibitively high information costs
(Darby & Karni, 1973; Anderson & Anderson, 1991). In this case, the credence
attributes of certain food-products can be identified via labeling. Studies have shown
that some credence attributes have positive impacts on consumer preference which
means that consumers are willing to pay a premium for certain quality attributes as
claimed by food-product labels (Burton et al., 2001; Umberger et al., 2002; Loureiro &
Umberger, 2003; Wirth et al., 2007; Dentoni et al., 2009; Gao et al., 2010).
In the US fresh produce industry, country of origin labeling (COOL) has been a
hot topic since COOL provisions for fresh fruits and vegetables first were included in the
24
Farm Security and Rural Investment Act (FSRIA) of 2002. FSRIA requires market
retailers, such as full-line grocery stores, supermarkets, and club warehouse stores, to
voluntarily label their products with COOL information stickers for consumers at the final
point of purchase. The USDA issued guidelines for voluntary country of origin labeling
(VCOOL) in 2002, which applied to the FSRIA covered products, including fresh fruits
and vegetables (VanSickle, 2008). In contrast, the 2002 US Farm Bill prescribed
mandatory country of origin labeling (MCOOL) instead of VCOOL. MCOOL, however,
was postponed due to the debate on its costs and benefits until the United States
Department of Agriculture’s final rule for MCOOL went into effect in March of 2009.
One of the issues that resulted from the change from VCOOL to MCOOL is
relative to consumers’ WTP for COOL. Much research has studied the effects of COOL
on consumer preference and WTP. For example, in 1965, Schooler was the first to
study the influence of COOL on consumers’ acceptance of products via empirical tests.
The results of the study indicated that consumers in the Central American Common
Market attributed certain characteristics to products from other member countries on the
basis of the country of origin. Since then, more research has been conducted on
consumer preferences regarding COOL for agricultural products, with most of the
research confirming that consumers prefer foods produced in their own country or
region (Verlegh & Steenkamp, 1999; Loureiro & Umberger, 2003, 2007 Chambers et al.,
2009).
In 2001, Schupp and Gillespie found that an average of 90.3% consumers in
Louisiana supported MCOOL for beef. Using data from a mail survey, they estimated a
probit model, and concluded that food safety concern was a significant factor in
25
increasing the probability of a consumer supporting MCOOL. Another finding was that
consumer preference for locally produced beef also positively affected the likelihood of
supporting MCOOL.
Similarly, Loureiro and Umberger (2003) showed that Colorado consumers were
willing to pay $1.53 and $0.70 more per pound for steak and hamburger produced in the
United States with a “US certified beef” label, respectively. They also found
demographic differences in WTP with female consumers most likely to pay a premium
for COOL and to be more supportive of MCOOL. Consumers with higher education
levels and higher household income were less willing to pay a premium for “US certified
beef”.
A nationwide survey conducted by Loureiro and Umberger (2004) showed that
consumers were willing to pay a relatively small price premium for US produced meat
due to their concerns about food safety issues associated with imported meat products.
They also found that consumers expressed more concerns about meat safety
inspections, production quality labels, and traceability of beef than they were about
COOL information.
Mabiso et al. (2005) used an experimental auction to estimate the predominant
determinants of WTP for COOL via a double-hurdle probit model for fresh apples and
tomatoes. The authors concluded that consumers were willing to pay a premium for
COOL. In the case of apples, 79% of the respondents were willing to pay a premium for
COOL, valued at $0.48 on average. Results were similar for tomatoes, with 72% willing
to pay an average of $0.49 as a premium.
26
Other research has identified various reasons why COOL information affects
consumer preference. Researchers have analyzed how COOL may help to signal or
suggest a specific degree of safety and/or quality regarding particular food-products
(Lewis & Grebitus, 2013). Verlegh and Steenkamp (1999) found that “country of origin
has a larger effect on perceived quality than on attitude toward the product or purchase
intention.” Becker et al. (2000) discovered that COOL was used by consumers to infer
and predict eating quality and safety. Loureiro and Umberger (2007) concluded that
COOL might only work as a signal of enhanced quality and safety given the fact that
consumers regarded foods with COOL information as having higher food quality and
better safety than foods without COOL information.
Consumers also may be willing to pay more for food products labeled with
country of origin information for emotional reasons. Verlegh and Steenkamp (1999)
concluded that there are three main mechanisms for COOL effects: cognitive, affective,
and normative cues. Cognitive cues refer to the concept that consumers use COOL to
infer the quality of the food. Affective cues refer to the emotional value a product has for
consumers, such as improving their social status (Batra et al., 2000). The normative cue
involves consumers’ social and personal norms relating to country of origin (Verlegh &
Steenkamp, 1999). For instance, preferring domestic food products is related to
negative attitudes toward foreign countries (Shimp & Sharma, 1987; Klein et al., 1998).
In addition to COOL research, “locally grown” food (local foods) has also been
studied in recent years. The factors that contributed to the increasing consumer
preference for local foods relative to out-of-state and imported foods have been
investigated by many researchers (Eastwood et al., 1987; Lehman et al., 1998; Brown,
27
2003; Loureiro et al., 2003; Campbell et al., 2004; Carpio & Isengildina-Massa, 2009;
Curtis et al., 2014). Similar to COOL, the most cited reason for consumer preference for
local foods is safety, health, and quality. Furthermore, consumers may have an altruistic
desire to support their community. They also may relate local production to being more
environmentally sustainable. Consumer preference for local food is enhanced when the
local food products are labeled as locally grown or state certified.
In summary, plenty of research has shown that COOL may impact consumers’
preferences and final purchasing decisions. Consumers were found to be more willing
to pay premiums on COOL for certain food products and to show a preference for
locally grown foods.
In the following sections of this thesis, chapter 3 analyzes cost effectiveness and
risk efficiency of MBr and its alternatives based on scientific trials conducted by the
University of Florida in 2013. Partial budgeting analysis and stochastic dominance
methods are used to compare all fumigations in the field trials and to identify the optimal
fumigation strategy. In chapter 4, the effect of three market COOL strategies on
consumer purchases of Florida/US versus Mexico tomatoes are analyzed based on a
mall survey conducted in 2014. Other factors such as demographics that may have an
effect on consumer preference and WTP are identified and estimated. Ordinal logistic
regression models and simple linear regression models are performed to determine the
impact of COOL and other factors that may have a significant influence on consumer
demand and willingness to pay for fresh tomatoes.
28
CHAPTER 3 ECONOMIC ANALYSIS OF FUMIGATION ALTERNATIVES, THE METHYL BROMIDE
BAN, AND ITS IMPLICATION
This chapter focuses on methyl bromide (MBr) phase-out and MBr alternatives in
the Florida tomato industry. As discussed in chapter 1, the US tomato industry has been
threatened by increasing production costs and decreasing yields resulting from the MBr
phase-out under the Montreal Protocol. As of 2014, scientific research has not found
feasible fumigant alternatives with the consistent, high technical effectiveness and low
cost of MBr. Alternative fumigants currently being used in the US tomato industry have
exhibited significant variations under different environmental conditions (Duniway, 2002;
Noling et al., 2005; Rosskopf et al., 2005; Noling et al., 2013). Therefore, this chapter
evaluates both the cost effectiveness and risk efficiency of MBr and the alternative
fumigants in the Florida tomato industry using partial budgeting analysis and stochastic
dominance methods. The findings will provide tomato growers and policy makers with
scientific information on optimal fumigation strategies to help reduce the production
costs and risk and boost yields of the struggling US tomato industry.
Fumigation Information in the Florida Tomato Industry
In the early transition period between 2005 and 2008 from MBr to alternative
fumigants, most growers used available stocks of MBr as the soil fumigation in their
crop production. Once MBr shortages began occurring in 2008, the industry had to find
effective substitutes for MBr. However, the use of alternative fumigants came with
issues of both technical and economic effectiveness. The FFVA survey in 2011/12 was
designed to determine alternative fumigation usage information to detect pests, weeds,
and disease problems associated with alternative fumigants and what additional
29
Integrated Cropping Practices were being implemented, and to estimate the production
losses due to repeated usage of MBr alternative fumigation treatments.
The 2011/12 FFVA survey interviewed 12 major tomato growers in Florida. The
total planted acreage of fresh tomato reported in the survey was about 8335 acres.
According to the survey, in 2010, tomato growers transferred from MBr:Pic (50:50) (50%
of Methyl bromide and 50% of Chloropicrin) to PicChlor 60 (1,3-Dichloropropene and
Chloropicrin). From 2010 to 2012, PicChlor 60 usage experienced a significant increase
while MBr:Pic (50:50) usage decreased. Due to its low cost and availability, PicChlor 60
became the most popular fumigant for Florida tomato growers. Between 2008 and
2012, only a few growers applied small amounts of other fumigants such as Midas
(98:2), Midas (50:50), and PaladinTM in their tomato fields.1
All tomato growers in the survey specified that MBr: Pic (98:2) or MBr:Pic (67:33)
had been used at an earlier point, as a fumigant treatment due to its higher efficacy in
pest, weed, and disease control. Growers also specified that current alternative
fumigation systems could cause up to a 20% tomato yield loss compared to MBr
fumigant treatments.
Based on the use of current MBr alternative fumigation systems, most growers
reduced or stopped double-cropping due to MBr alternative/substitutive fumigations
failing to control pests, weeds or diseases as effectively as MBr. As a result, double-
cropping yields dropped significantly.
1 In 2008, Midas was granted a conditional registration for Florida but was suspended in 2012, and PaladinTM was only available for limited use in Florida under an experimental use permit (EUP), with full registrations pending (Rosskopf et al., 2010).
30
The survey results also indicated that integrated cropping practices, which refer
to the additional practices utilized to minimize pests or weeds or to control a component
of the soil-borne pest complex, were needed in commercial tomato production under the
current fumigation systems. Specifically, most growers applied glyphosate, Roundup,
and/or Sandea as additional herbicides in summer fallow. In addition, growers had to
perform extra disking and hand weed pulling which increase labor costs; some growers
even planted primary cover crops such as sorghum to manage soil fertility and quality.
Because current MBr alternatives have performed poorly compared to MBr
treatments in controlling nematodes, weeds, and diseases, there has been more crop
yield losses. Nutsedge and Fusarium wilts were identified by tomato growers as
affecting yield the most, causing 10% to 20% yield losses. PicChlor 60 has failed to
provide adequate technical efficacy against nutsdges and Fusarium wilts.
In summary, the FFVA survey revealed that the current MBr alternative
fumigation treatments in Florida tomato production have been less effective than MBr
fumigation treatments, leading to lower crop yields, higher production costs, and
increased financial losses for the growers.
Materials and Methods
Two analytical tools are employed to analyze the cost effectiveness and risk
efficiency of MBr fumigation treatments versus the MBr alternative fumigation treatment
systems, including non-fumigated treatment, used in the field trials conducted by the
University of Florida in Balm, Florida in fall 2013. Partial budgeting analysis is employed
to evaluate and analyze the cost effectiveness and economic efficacy of MBr treatments
and its substitutes. Stochastic dominance approaches, including second-order
stochastic dominance (SSD) and stochastic efficiency with respect to a function (SERF),
31
are used to identify and rank different fumigation strategies in the field trials based on
risk efficiency under given risk aversion.
Partial Budgeting Analysis
Partial budgeting analysis is a standard technique to assess the economic impact
of a change in a farm system (Kay et al., 1994). This type of analysis is frequently used
to estimate the impact of various alternative production techniques when the changes
involve only parts of the production system (Warmann, 1995; Roberts & Swinton, 1996;
Wossink & Osmond, 2002). The partial budgeting technique compares the negative
effects of applying a new treatment relative to a base or standard treatment to the
positive effects associated with the new treatment (Sydorovych et al., 2008). Negative
effects consist of added costs and reduced returns of the alternative treatment
compared with the base treatment; while positive effects include reduced costs and
added returns of the alternative treatment compared with the base treatment. In this
case, average yields, fumigation costs, harvest & marketing costs, and net returns that
change with different fumigation treatments should be considered. Those costs that are
fixed across treatments are excluded in partial budgeting analysis. It is assumed that
the added/reduced costs for the alternative treatment were incurred if the alternative
treatment caused higher/lower fumigation costs or it resulted in higher/lower harvest &
marketing costs. In this analysis, the base treatment is MBr:Pic (67:33), the MBr
treatment most commonly used by tomato growers in Florida before the MBr ban.
Stochastic Dominance and Stochastic Efficiency Analysis
Stochastic dominance (SD) is a useful economic tool to analyze risk returns of
different choices (e.g., investment portfolios or treatments) through stochastic ordering.
Stochastic dominance can act as a valid efficiency criterion to determine the risk
32
efficient set of various alternatives of input factors in production such as fertilizers,
fumigants, and pesticides. Simetar®, a statistical software developed by Texas A&M
University, which focuses on risk analysis and management in agribusiness, is used to
perform the stochastic dominance analysis.
Using the SD approach, both the performance of different fumigations and their
consistency under different conditions are considered. Second-order stochastic
dominance (SSD) is performed to analyze the risk and returns of these alternatives to
determine the risk efficient set of various fumigation treatments. The SSD analysis
incorporates risk aversion, assuming the utility function is concave; that is, the second
derivative of the utility function is negative, u’’<0. This eliminates certain dominated
distributions from the efficiency set of first-order stochastic dominance (FSD) through
restricting risk aversion. Mathematically, a necessary and sufficient condition for an
option F to dominate an alternative G by all risk-averse decision makers is:
∫𝐹(𝑧)𝑑𝑧 ≤
𝑥
−∞
∫𝐺(𝑧)𝑑𝑧
𝑥
−∞
for all possible x (Hadar & Russell, 1969).
Graphically, a distribution is considered dominant if it “lies more to the right in
terms of differences in area between the cumulative distribution function (CDF) curves
cumulated from the lower values of the uncertain quantity (Anderson et al., 1977).”
Furthermore, stochastic efficiency with respect to a function (SERF) developed
from SSD is also conducted in the analysis. The SERF method orders a set of risky
alternatives in terms of certainty equivalents calculated for specified ranges of risk
aversion coefficients (Hardaker et al., 2004). The certainty equivalent (CE) is equal to
33
the amount of a certain payoff to which a decision maker would be indifferent when
compared with a risky investment or treatment. Alternatives with higher CE values are
more preferable. The CE value of a risk-averse decision maker is typically less than the
expected money value (EMV) of the risky alternative, and the difference between EMV
and CE is considered as the risk premium (RP). The RP represents the minimum
amount that has to be paid to a risk-averse decision maker so that he/she is willing to
switch from the less risky alternative to the more risky alternative (Hardaker et al.,
2004). Figure 3-1 illustrates the concepts of CE and RP. In this figure, 𝑊 is wealth, 𝑈 is
utility, 𝐸(𝑢) is the expected utility, 𝐸(𝑤) is the expected money value, 𝐶(𝑤) is the CE,
and the difference between 𝐸(𝑤) and 𝐶(𝑤) is the RP 𝜋(𝑤).
SERF divides the range of the risk-aversion coefficients (RACs) into equal
intervals so that it can evaluate all CEs for all risky alternatives at every interval
simultaneously through different utility functions (Hardaker et al., 2004).2 In SERF, the
negative exponential utility function is the default under constant absolute risk aversion
(CARA). Hardaker et al. (2004) indicated that the negative exponential utility function in
SERF accords with the hypothesis that producers prefer less risk to more risk given the
same expected return, and it assumes that decision makers have constant absolute risk
aversion. Babcock et al. (1993) and Schumann et al. (2004) noted and demonstrated
that negative exponential utility function can be used reasonably to analyze a grower’s
decision under certain risk.
However, under CARA utility function, it implies that the changing initial wealth w
does not affect the economic decision, namely that CARA implies zero “wealth effects”.
2 Simetar® specifies 23 equal intervals.
34
Although CARA is a convenient assumption, some empirical research may find it more
plausible that ARA is decreasing with wealth level (DARA), meaning that if the wealth of
a decision maker increases, then he/she will choose to increase the number of dollars
of the risky asset held in the portfolio. Simply put, richer people take higher risks. In this
case, Constant Relative Risk Aversion (CRRA) is more plausible than CARA, which
takes the form as follows:
𝑈(𝑊) = {
𝑊1−𝑟
1 − 𝑟 𝑖𝑓 𝑟 > 0, 𝑟 ≠ 1
𝐿𝑛(𝑊) 𝑖𝑓 𝑟 = 1
where r is the relative risk-averse coefficient (RRAC).
Under the power utility function, risk aversion and CRRA always imply
Decreasing Absolute Risk Aversion (DARA). Specifically, CRRA means that when the
wealth of the decision maker increases, he/she will increase the level of risky assets but
the proportion of risky assets in his/her total assets does not change. In this study, the
CRRA power utility function is used in SERF to calculate the CE. The range of RRAC is
set from 1 to 4, representing “normal risk averse” to “extremely risk averse” (Anderson &
Dillon, 1992).
Specifically, the formula for calculating CE with power utility function is (Clemen,
1991):
𝐸(𝑈) =∑𝑝𝑖(𝑋𝑖 + 𝜔)
(1−𝑟)
1 − 𝑟𝑖
𝐶𝐸 = [𝐸(𝑈)(1 − 𝑟)]11−𝑟 − 𝜔
where
𝐶𝐸 = 𝑐𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 𝑒𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡
35
𝐸(𝑈) = 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑢𝑡𝑖𝑙𝑖𝑡𝑦
𝑟 = 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑟𝑖𝑠𝑘 𝑎𝑣𝑒𝑟𝑠𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡
𝑋𝑖 = 𝑟𝑖𝑠𝑘 𝑜𝑢𝑡𝑐𝑜𝑚𝑒
𝑝𝑖 = 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑋𝑖
𝜔 = 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑤𝑒𝑎𝑙𝑡ℎ
In the SERF analysis, the curves of the CEs of all the risky alternatives are
graphed on the vertical axis against risk aversion on the horizontal axis over the RRAC
range. The SERF graph facilitates the presentation of ordinal rankings for decision
makers with different risk attitudes and provides a measure of decision makers’
preferences among different risky alternatives at each risk aversion level (Hardaker et
al., 2004).
In this study, a representative farm is assumed in the SERF analysis which
represents a tomato farm in Florida with average farm equity and farm size. The farm
equity is regarded as the initial wealth of the farm. Based on the published farm financial
data from the Agricultural Resource Management Survey of the United States
Department of Agriculture (USDA/ARMS, 2013), the average farm equity of vegetable
farms in Florida is about $3,420,056. Then according to the Census of Agriculture
conducted by the National Agricultural Statistics Service of the United States
Department of Agriculture (USDA/NASS, 2013c), the average tomato farm size in
Florida is about 70 acres. Therefore, the representative farm analyzed in SERF is a
tomato farm in Florida with 70 acres land and $3,420,056 farm equity.
36
Data Description and Adjustment
Data Description
Yield and input use data were collected from field trials conducted by the
University of Florida in fall 2013. The researchers used two MBr fumigant treatments,
MBr:Pic (67:33) (67% of Methyl bromide and 33% of Chloropicrin) and MBr:Pic (50:50)
(50% of Methyl bromide and 50% of Chloropicrin); three alternative fumigant treatments,
including TE-3 (Telone II, Chloropicrin, and DMDS), PicChlor 60 (1,3-Dichloropropene
and Chloropicrin), and FL-3 way (Telone II, Chloropicrin and K-pam); and a non-
fumigated treatment. There were four fields and each field had four replicate blocks.
Each block had three beds, divided into six plots, and each plot was 2.67 feet wide by
75 feet long (200 square feet = 0.0046 treated acre). Six treatments were applied on the
six plots, respectively, in each block.
The four fields had different levels of weed and nematode populations in order to
test the efficacy of MBr and its fumigant alternatives. The application rates of each
fumigant treatment for the four fields were as follows: 350 lbs/acre of MBr:Pic (67:33),
350 lbs/acre of MBr:Pic (50:50), 400 lbs/acre of TE-3, and 300 lbs/acre of PicChlor 60
and FL-3 way (122 lbs/acre of Telone II, 150 lbs/acre of Chloropicrin, and 60 gals/acre
K-pam). All treatments were covered with virtually impermeable film (VIF) mulch and
drip tapes, and applied the same amount of fertilizers, pesticides, fungicides, and
irrigation were applied to each field during the growing season. Unlike commercial
operations, herbicides were not applied in the field trials.
Fumigation costs consisted of costs for materials, machinery, and labor. For
fumigant material costs, the market prices of MBr:Pic (67:33) and MBr:Pic (50:50) used
in field trials were $12.00/lb and $8.00/lb, respectively, which were much higher than
37
other fumigant prices. This was because restrictions on MBr production, import, and
consumption under the Montreal Protocol have driven the market price of MBr up in the
United States in recent years. In order to make a justifiable analysis, the original price of
MBr before the ban in 1997 (adjusted for inflation) was used. According to production
budget statistics from the FFVA, the price of MBr in 1997 in Florida tomato production
areas was about $3.01/lb. After being adjusted with the Producer Price Index (PPI) for
pesticides, fertilizers, and other agricultural chemicals released by the United States
Bureau of Labor Statistics, the MBr price in 2013 was reset at $5.45/lb (USBLS, 2013).
Accordingly, the MBr:Pic (67:33) and MBr:Pic (50:50) prices were changed to $4.80/lb
and $4.46/lb, respectively. In addition, VIF mulch and drip tape costs were also
incorporated into the fumigation material costs, estimated at $448/acre and $150/acre,
respectively.
Fumigation machinery costs only included fuel and lubrication costs.
Depreciation and other non-cash overhead were assumed to remain unchanged across
treatments and therefore excluded. For each fumigation treatment, per plot tractor time
was recorded. Then according to tractor fuel consumption information from test reports
published by the Nebraska Tractor Test Laboratory (2013), the fuel cost was estimated
by multiplying average tractor time by per hour fuel consumption based standard power
takeoff (PTO) horsepower. Lubricant cost was assumed as 10% of the fuel cost, as
suggested by a 2007 study on sample costs to produce fresh market tomatoes
conducted by the University of California at Davis (Stoddard et al., 2007).
For the fumigation labor costs, it was assumed that the fumigation operation
required one tractor driver and three field workers to lay VIF mulch and drip tapes.
38
Based on farm labor rate information released by the National Agricultural Statistics
Service of the United States Department of Agriculture (USDA/NASS, 2013b), the
hourly wage rates during the July 2013 reference week were $12.55 for all hired
workers and $10.70 for field workers. Labor time for tractor driver was set to be 20%
higher than the average operation time to account for the extra labor for activities such
as equipment setting up, moving, maintenance, and field repair. After calculating the
average labor operation time for each fumigation treatment, the labor costs were
estimated by multiplying the unit wage rate by average labor hours.
Harvest & marketing costs included pick/pack/haul cost, container cost, selling
cost and organization fee. Using “Estimated tomato production costs in the
Manatee/Ruskin area” published by UF in 2009 as the reference (UF/IATPC, 2009), the
harvest & marketing costs in this study were estimated at $3.51/carton after being
adjusted for inflation (each carton contained approximately 25 lbs of tomatoes).3 Except
for fumigation costs, herbicide costs, and harvest & marketing costs, other cost items in
this reference were used after the adjustment for inflation.4
Tomato market prices used in the analysis were acquired from the Agricultural
Market Service of the United States Department of Agriculture (USDA/AMS, 2013). The
harvested tomatoes were graded and met the requirement for US No.1, US
Combination, or US No.2 of the US Standard for Grades Tomatoes. The three standard
3 The estimated costs in this reference were structured as “pre-harvest cost” and “harvest and marketing costs”. The pre-harvest cost included 1) operating costs, which covered major variable cost items such as fertilizer, fumigants, herbicides, 2) miscellaneous, which included minor variable cost items such as scouting and stakes, 3) fixed costs, including land rent, machinery fixed cost, and farm management and overhead.
4 The factor of 0.822, derived based on the Producer Price Index (PPI) of tomato production published by the Bureau of Labor Statistics, was used to adjust the costs.
39
graded tomatoes had respective wholesale prices. Average wholesale prices of fresh
tomatoes in southwest Florida in 2013 were used in the analysis.
The net return was calculated using gross revenue minus fumigation costs,
harvest & marketing costs, and other costs, excluding herbicide costs. Besides net
return, gross margin was also calculated using gross revenue minus fumigation costs,
harvests costs, and other variable costs, excluding fixed costs.
Data Adjustment
In total, 96 yield observations were collected and each treatment had 16 yield
observations. Except for treatments, the experiment environment and other inputs were
fixed across the fields. However, since the four fields were not identical due to different
levels of nematodes and weeds, field discrepancy also affected tomato yields. In this
case, it was assumed that there were only two factors (treatment and field) affecting the
tomato yields. A linear regression model was established to separate the treatment
effects on yield from the field effects. Dummy variables were created for six treatments
and four fields. The regression function was specified as follows:
𝑌𝐿 = 𝑎 + 𝛽2𝑇𝑅𝑇2 + 𝛽3𝑇𝑅𝑇3 + 𝛽4𝑇𝑅𝑇4+ 𝛽5 𝑇𝑅𝑇5 + 𝛽6TRT6 + 𝛾1𝐹𝐿1 + 𝛾2𝐹𝐿2 +
𝛾3𝐹𝐿3 + 𝑒
where 𝑌𝐿 represents yield, 𝑇𝑅𝑇 represents treatment, and 𝐹𝐿 represents field, and. 𝑒 is
the error term.
TRT2 is MBr:Pic (67:33), TRT3 is MBr:Pic (50:50), TRT4 is TE-3, TRT5 is
PicChlor 60, and TRT6 is FL-3 way. Since dummy variables are used in the regression,
to avoid the dummy variable trap, a benchmark category is needed; therefore the
intercept stands for the combination of TRT1 & FL4, namely the non-fumigated
40
treatment in Field No. 4. The error term includes all other potential risk factors that can
impact tomato yields besides the two factors of treatment and field.
The regression model is estimated using the ordinary least square (OLS)
method. Regression results (Table 3-1) illustrate that except for TRT6 (the FL-3 way),
other fumigation treatments do have positive effects on tomato yield if compared to the
non-fumigated treatment in Field No. 4, and are significant at the 5% level. All field
variables are significant statistically at the 5% level. Specifically, TRT2, namely MBr:Pic
(67:33), has the maximum coefficient, estimated at 61.119, followed by TRT3, MBr:Pic
(50:50) estimated at 52.663; TRT4, TE-3 estimated at 44.038; and TRT5, PicChlor 60
estimated at 25.150. As for field variables, FL1 and FL2 have positive effects on yields,
but FL3 has a negative effect on yield compared to Field No. 4. The estimated
parameters for FL1, FL2, and FL3 are 48.448, 41.371, and –40.188, respectively. Since
the field effects have been determined, in the following analyses, the yield data is
adjusted to remove field effects.
Results
Fumigation Costs of MBr and Alternative Soil Treatments
Table 3-2 indicates the costs of fumigation with MBr:Pic (67:33) and its
alternative treatments. Estimated fumigation costs vary over treatments. Specifically,
fumigation costs are decomposed into material costs, labor costs, and machinery costs.
The projected fumigation costs of the base treatment, MBr:Pic (67:33), are the most
expensive, estimated at $2391.94/acre. Other alternative soil treatments and the non-
fumigated treatment result in savings in fumigation costs relative to MBr:Pic (67:33).
The non-fumigated treatment is the least expensive alternative at $711.78/acre for
fumigation costs, including the estimates of machinery, material, and labor costs of
41
laying the VIF mulch and drip tapes. The differences in fumigation costs for each
alternative treatment relative to MBr:Pic (67:33) are also listed in Table 3-2. Other than
the non-fumigated treatment, the most cost-savings fumigation treatment is PicChlor 60,
followed by TE-3, FL-3 way, and MBr:Pic (50:50).
Estimated Yields, Total Costs, and Net Returns Associated with MBr and Its Alternative Treatments
Average tomato yields of MBr:Pic (67:33) and its selected alternative treatments
are shown in Table 3-3. The results show that the highest average marketable yield is
associated with MBr:Pic (67:33), the base treatment in the field trials, which is 31402.17
lbs/acre. Another MBr treatment, MBr:Pic (50:50), produces the second highest average
marketable yield (39563.86 lbs/acre), followed by TE-3 (27688.86 lbs/acre), PicChlor 60
(23582.88 lbs/acre), and FL-3 way (21334.24 lbs/acre). The non-fumigated treatment
has the lowest average yield of 18115.49 lbs/acre. Because unit harvest & marketing
cost is estimated at $3.51/25-lb carton as discussed above, total harvest & marketing
cost for each fumigation treatment is in direct proportion to its average tomato yield.
Average wholesale prices of fresh tomatoes in southwest Florida in 2013 were
$12.65/carton for jumbo and extra-large tomatoes, $11.21/carton for large tomatoes,
and $10.50/carton for medium and small tomatoes. Projected gross revenues were
estimated using tomato market prices and average yields graded in different fruit size
categories. The base treatment, MBr:Pic (67:33), leads to the highest average gross
revenue at $14484.39/acre. TE-3 performs the best among the selected alternative
fumigation treatments except for the two MBr treatments. Its estimated gross revenue is
about $12735.98/acre and the gross revenue losses relative to both of the MBr
treatments are less than those for PicChlor 60, FL-3 way, and the non-fumigated
42
treatments. The fumigation currently most used in Florida tomato production, PicChlor
60, would incur a loss of $3568.52/acre in gross revenue relative to MBr:Pic (67:33). FL-
3 way treatment results in more losses in gross revenue relative to MBr:Pic (67:33),
estimated at $4402.54/acre. The non-fumigated treatment causes the greatest gross
revenue loss ($6175.20/acre) relative to MBr:Pic (67:33).
When it comes to average net return, all treatments result in negative values.
According to Table 3-4, base treatment MBr:Pic (67:33) leads to the lowest average
loss in net return at –$749.32/acre, followed by MBr:Pic (50:50) (–$1236.15/acre), TE-3
(–$1605.15/acre), and PicChlor 60 (–$2420.04/acre). FL-3 way incurs the highest
average loss in net return, reaching –$3559.02/acre, even higher than the non-
fumigated treatment (–$3378.91). All other fumigation treatments generate positive
gross margin except for FL-3 way (–$369.43/acre) and the non-fumigated treatment (–
$189.32/acre). MBr:Pic (67:33) has the highest gross margin, reaching $2440.27/acre
on average, followed by MBr:Pic (50:50) ($1953.44/acre), TE-3 ($1584.44/acre), and
PicChlor 60 ($769.55/acre).
Table 3-4 also indicates the average breakeven grower price under each
treatment required to cover total fresh tomato production costs. The breakeven grower
price for MBr:Pic (67:33) is about $12.13/carton, followed by MBr:Pic (50:50)
($12.57/carton), TE-3 ($12.95/carton), PicChlor 60 ($14.14/carton), FL-3 way
($15.98/carton), and the non-fumigated treatment ($16.13/carton).
Table 3-5 displays the negative and positive effects of each fumigation treatment
relative to the base treatment, calculated with the estimated fumigation costs (Table 3-
2), and harvested costs and gross revenues (Table 3-3). Except for fumigation and
43
harvest & marketing costs, other costs in production remained unchanged across the
fumigation treatment applications so these costs (e.g., fertilizer, pesticide, fungicide,
etc.) are excluded from the analysis. The total negative effects of one alternative
treatment consisted of the added costs and reduced returns relative to the base
treatment, and total positive effects included reduced costs and added returns. The total
net effect of one alternative treatment relative to the base treatment is the difference in
the value of its positive and negative effects.
MBr:Pic (67:33) is the base treatment so its total effect is zero. It has the highest
fumigation costs and harvest & marketing costs so no extra costs are added into the
alternative treatment relative to MBr:Pic (67:33). The non-fumigated treatment has the
largest negative effects relative to MBr:Pic (67:33) due to its worst yield performance,
reaching –$6175.20/acre, followed by FL-3 way, PicChlor 60, TE-3 and MBr:Pic (50:50),
which are estimated at –$4402.54/acre, –$3568.52/acre, –$1748.42/acre and –
$856.42/acre, respectively. As for total net effects of the alternative treatments relative
to MBr:Pic (67:33), all the alternatives lead to negative total net effects; FL-3 way has
the largest negative net effects at –$2809.70/acre due to its high fumigation costs and
poor yield performance, followed by non-fumigated treatment, PicChlor 60, TE-3 and
MBr:Pic (50:50), estimated at –$2629.59/acre, –$1670.72, –$855.827, and –
$486.33/acre, respectively. It can be concluded that although fumigation costs of the
MBr:Pic (67:33) treatment are the highest among all the treatments, its outstanding yield
performance still makes it the most cost-effective fumigation, producing more positive
effects than the other treatments. The results also reflect that TE-3 performs the best
among the MBr alternatives.
44
Stochastic Dominance and Stochastic Efficiency Analysis Results
Adjusted tomato yields are considered in the SSD analysis and the net returns of
the different fumigation treatments and the non-fumigated treatment are studied through
SERF analysis. The SSD result (Table 3-6) generalized from the CDF graph (Figure 3-
2) shows that the two MBr treatments, MBr:Pic (67:33) and MBr:Pic (50:50), dominate
the other three alternative fumigants and the non-fumigated treatment; likewise, TE-3
dominates PicChlor 60 and FL-3 way.
In the SERF analysis, as discussed above, the RRAC range in the CRRA power
utility function is set from 1 to 4, representing low-risk aversion to high-risk aversion.
The summary statistics of the net returns and initial wealth associated with each
treatment is shown in Table 3-7.
The SERF results are plotted across RRAC values by treatment in Figure 3-3.
The results illustrate that MBr:Pic (67:33) produces the highest CE values, followed by
MBr:Pic (50:50), TE-3, PicChlor 60, FL-3 way, and the non-fumigated treatment within
the given RRAC range. FL-3 way and the non-fumigated treatment intersect at the
breakeven RRAC (about 1.30) but after that, as RRAC increases, FL-3 way has a
higher CE value than does the non-fumigated treatment. It can be concluded that the
most risk-efficient fumigation strategy is MBr:Pic (67:33), which has the highest CE
values. TE-3 has the highest CE values among all MBr alternatives, better than
PicChlor 60, the most popular fumigation currently used in the Florida tomato industry.
The RPs associated with fumigation treatments relative to the non-fumigated
treatment are mapped across the RRAC values in Figure 3-4. The RPs are largest for
MBr:Pic (67:33), reaching an average $2509.43/acre across the RRAC range, followed
by MBr:Pic (50:50) ($2193.63/acre), TE-3 ($1349.40/acre), PicChlor 60 ($994.71/acre),
45
and FL-3 way ($154.71/acre). Before the breakeven RRAC point, RPs for FL-3 way are
negative. As the RRAC increases, the RPs demanded for MBr:Pic (67:33) and TE-3
decrease while the RPs for MBr:Pic (50:50), PicChlor 60, and FL-3 way increase. This
is because the larger standard deviation of the net return associated with a treatment
means the net return is more risky and unstable. Therefore, as the decision maker
becomes more risk averse, he/she prefers less risky treatments so the RPs for these
treatments should increase.
46
Table 3-1. Regression results on tomato yield adjustments from SAS
Variable Parameter Estimate
Standard Error t Value Pr > |t|
Intercept 70.923 8.581 8.27 0.0001*** TRT2 61.119 9.908 6.17 0.0001*** TRT3 52.663 9.908 5.32 0.0001*** TRT4 44.038 9.908 4.44 0.0001*** TRT5 25.150 9.908 2.54 0.0129** TRT6 14.803 9.908 1.49 0.1388 FL1 48.448 8.090 5.99 0.0001*** FL2 41.371 8.090 5.11 0.0001*** FL3 –40.188 8.090 –4.97 0.0001***
Number of Observation Used: n=96. Adj R-Sq: 0.6814. “*”: Significant at 10% “**”: Significant at 5% “***”: Significant at 1% Table 3-2. Estimated average tomato fumigation costs for MBr:Pic (67:33) and selected alternative soil treatments and the fumigation costs of the alternative treatments relative to MBr:Pic (67:33)
MBr and selected
alternative treatment
Fumigation labor costs
($/acre)
Fumigation machinery
costs ($/acre)
Fumigation materials
costs ($/acre)
Total fumigation
costs ($/acre)
Fumigation costs relative
to MBr:Pic (67:33) ($/acre)
Non-fumigated 70.01 43.77 598.00 711.78 –1680.16 MBr:Pic (67:33) 64.79 49.15 2278.00 2391.94 0.00 MBr:Pic (50:50) 69.06 52.39 2159.00 2280.45 –111.49
TE-3 69.77 52.39 1898.00 2020.71 –371.24 PicChlor 60 73.33 55.63 1463.00 1591.97 –799.97 FL-3 way 92.99 106.95 2012.70 2212.64 –179.30
Note: The fumigation machinery costs only included diesel and lubricant costs; depreciation and other non-cash overhead were excluded.
47
Table 3-3. Marketable tomato yields, gross revenue for MBr:Pic (67:33) and selected alternative fumigant treatments, and the difference in gross revenues relative to MBr:Pic (67:33)
MBr and selected alternative treatment
Jumbo and extra-large tomato
(lbs/Acre)
Large tomato
(lbs/Acre)
Medium and small tomato
(lbs/Acre)
Total Yield (lbs/Acre)
Gross Revenue ($/acre)
Gross Revenues relative to
MBr:Pic (67:33) ($/acre)
Non-fumigated 6633.15 4605.98 6875.00 18115.49 8309.20 –6175.20 MBr:Pic (67:33) 12759.51 6977.58 11665.08 31402.17 14484.39 0.00 MBr:Pic (50:50) 11872.28 6694.97 10996.60 29563.86 13627.97 –856.42
TE-3 10855.30 6095.11 10738.45 27688.86 12735.98 –1748.42 PicChlor60 9974.86 5395.38 8212.64 23582.88 10915.88 –3568.52 FL-3 way 11502.72 4656.25 5175.27 21334.24 10081.85 –4402.54
Note: Tomato marketing prices are: $12.65/carton for jumbo and extra-large tomatoes, $11.21/carton for large tomatoes and $10.5/carton for medium and small tomatoes (Each carton contains approximately 25lb tomatoes). Table 3-4. Estimated average costs incurred to produce, fumigate, and harvest tomatoes, total net returns, and net returns of the selected alternative soil treatments relative to MBr:Pic (67:33)
MBr and selected alternative treatments
Fumigation cost ($/acre)
Harvest & marketing
cost ($/acre)
Total costs ($/acre)
Gross margin ($/acre)
Net return
($/acre)
Net returns relative to
MBr:Pic (67:33) ($/acre)
Break-even grower price
($/carton)
Non-fumigated 711.78 2543.41 11688.10 -189.32 -3378.91 -2629.59 16.13 MBr:Pic (67:33) 2391.94 4408.87 15233.72 2440.27 -749.32 0.00 12.13 MBr:Pic (50:50) 2280.45 4150.77 14864.13 1953.44 -1236.15 -486.83 12.57
TE-3 2020.71 3887.52 14341.13 1584.44 -1605.15 -855.83 12.95 PicChlor60 1591.97 3311.04 13335.91 769.56 -2420.04 -1670.71 14.14 FL-3 way 2212.64 2995.33 13640.88 -369.43 -3559.02 -2809.70 15.98
Note: Other than fumigation cost and harvest & marketing cost, other production costs for each treatment are estimated at $8432.91/acre and the fixed costs for each treatment are estimated at $3189.59/acre; total production costs exclude the herbicide costs since herbicides were not applied in the field trials.
48
Table 3-5. Total negative effects (added costs and reduced returns), total positive effects (reduced costs and added returns), and total effects of the selected alternative soil treatments relative to MBr:Pic (67:33) in the tomato production system
MBr:Pic (67:33) and its
alternatives
Added costs of the
alternative treatment ($/acre)
Reduced returns of the
alternative treatment ($/acre)
Total negative
effects of the
alternative treatment ($/acre)
Reduced costs of the alternative treatment ($/acre)
Added returns of
the alternative treatment ($/acre)
Total positive effects of the alternative treatment ($/acre)
Total effects
relative to MBr:Pic (67:33)
($/acre)
Non-fumigated 0 6175.20 6175.20 3545.61 0 3545.61 -2629.59 MBr:Pic (67:33) 0 0 0 0 0 0 0 MBr:Pic (50:50) 0 856.42 856.42 369.59 0 369.59 -486.83
TE-3 0 1748.42 1748.42 892.59 0 892.59 -855.827 PicChlor60 0 3568.52 3568.52 1897.80 0 1897.80 -1670.72 FL-3 way 0 4402.54 4402.54 1592.84 0 1592.84 -2809.70
49
Table 3-6. Second-order stochastic dominance result of adjusted yield of methyl bromide and its alternative treatments
Fumigant and Non-fumigated Treatments
Non-fumigated MBr:Pic (67:33)
MBr:Pic (50:50)
TE-3 PicChlor 60 FL-3 way
Non-fumigated SDD: MBr:Pic (67:33) SDD: Non-fumigated TE-3 PicChlor 60 FL-3 way MBr:Pic (50:50) SDD: Non-fumigated TE-3 PicChlor 60 FL-3 way
TE-3 SDD: PicChlor 60 FL-3 way PicChlor 60 SDD: FL-3 way SDD:
Table 3-7. Summary statistics of net returns of MBr:Pic (67:33) and its alternatives
Non-fumigated MBr:Pic (67:33)
MBr:Pic (50:50)
TE-3 PicChlor 60 FL-3
Average Net Returns -3378.33 -749.33 -1236.16 -1605.16 -2420.04 -3559.37 Standard Deviation 4424.96 4896.38 4158.88 6166.20 4373.97 2531.49
Minimum -9919.63 -10326.00 -8645.82 -10343.2 -9142.85 -9047.74 Maximum 2870.73 6734.38 3276.90 11296.66 4622.12 -342.12
Initial wealth=$3,420,056
Average farm size=70 acres
50
Figure 3-1. Explanatory figure of expected money value, certainty equivalent, and risk
premium
51
Figure 3-2. Comparison of MBr:Pic (67:33) and its alternatives’ CDF series for adjusted
yield data
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 50 100 150 200 250
Pro
bComparison of 6 CDF Series
Non-fumigated MBr:Pic (67:33) MBr:Pic (50:50)
TE-3 PicChlor 60 FL-3
52
Figure 3-3. Comparison of the SERF results of the six treatments’ net returns under
power utility function
Non-fumigated
MBr:Pic (67:33)
MBr:Pic (50:50)
TE-3
PicChlor 60
FL-3 way
-350,000.00
-300,000.00
-250,000.00
-200,000.00
-150,000.00
-100,000.00
-50,000.00
0.00
0 1 2 3 4 5
Cert
ain
ty E
qu
iva
len
t
RRAC
Stochastic Efficiency with Respect to A Function (SERF) Under a Power Utility Function
Non-fumigated MBr:Pic (67:33) MBr:Pic (50:50)
TE-3 PicChlor 60 FL-3 way
53
Figure 3-4. Comparison of the six fumigation treatments’ risk premiums under the
power utility function relative to the non-fumigated treatment ($/acre)
Non-fumigated
MBr:Pic (67:33)
MBr:Pic (50:50)
TE-3
PicChlor 60
FL-3 way
-500.00
-
500.00
1,000.00
1,500.00
2,000.00
2,500.00
3,000.00
0 1 2 3 4 5
RRAC
Power Utility Weighted Risk Premiums Relative to Non-fumigated
Non-fumigated MBr:Pic (67:33) MBr:Pic (50:50)
TE-3 PicChlor 60 FL-3 way
54
CHAPTER 4 DETERMINING THE IMPACT OF STATE-SPECIFIC SIGNS AND LABELS ON
TOMATO MARKETING
As discussed in chapter 1, the evolving market conditions and trade relationship
between the United States and Mexico is posing tremendous challenges to the Florida
tomato industry. Against such a background, this chapter provides the industry with
marketing information to understand consumer demand for local fresh tomatoes
(Florida/US).
Consumer surveys were conducted to investigate the effect of three market
labeling stimuli regarding origin of production. The three marketing label stimuli include
various labels comparing Florida or US production to Mexican production. The goal of
this research is to identify which, if any, of the labeling strategies would lead to
increased willingness to pay (WTP) from consumers for Florida/US tomatoes. The
contingent valuation method (CVM) is used to estimate the WTP toward different
labeled tomatoes among consumers. Logistic regression and linear regression models
are used to determine the factors influencing consumers’ behavior of reading tomato
COOL, consumer choices, and WTP for local tomatoes.
Survey Designing and Data Collection
To determine the impact of different types of information on production origin, a
consumer survey was designed and conducted using the mall intercept format. In this
way, how participants’ respond to displays regarding origin of production in a setting
that is similar to a grocery store can be observed cost-effectively.
The working hypothesis is that consumers are willing to pay a premium for
tomatoes from Florida/US compared to tomatoes from Mexico. It is further hypothesized
that consumers will respond differently to the origin information in each label scenario,
55
with the least noticing origin in the case of plain labels (current situation) with country of
origin, followed by plain labels with Florida identification, followed by the case with point-
of-purchase signage.
To test the hypothesis, a mall intercept questionnaire is conducted regarding the
surveyed consumers’ fruit and vegetable consumption, their fresh produce purchasing
habits, and their COOL information awareness, and the impact that different ways of
presenting origin information has on consumer choice of fresh tomatoes. Participants
are screened to meet the following criteria: an adult (older than 18 years old), primary
grocery shopper who purchased fresh tomatoes at least once/month in the past few
months. After answering the baseline questions on frequency and location of grocery
shopping, participants are presented with two wooden baskets of tomatoes in a setting
similar to what would be found in a real produce section of a store. The two baskets are
immediately next to each other on a table (as they would be on display in the store).
The tomatoes in the two baskets are exactly the same type of tomatoes except for the
different label information that indicates production origins. Random 3-digit numbers are
assigned to each label scenario to enable consumers to respond to questions about the
tomatoes without calling specific attention to the labels. The origin information is labeled
and presented in three different formats: 1) US and Mexico stickers on tomatoes
(tomato #599 vs. tomato #280); 2) Florida and Mexico stickers on tomatoes (tomato
#462 vs. tomato #280); and 3) “Grown in Florida” sign on top of the basket (similar to
point-of-purchase information in a store) plus US stickers and Mexico stickers on
tomatoes (tomato #828 vs. tomato #280). Tomato #599, tomato #462, and tomato #828
56
all refer to Florida tomatoes but with different COOL strategies, namely Florida/US
tomatoes.
Participants are randomly assigned to one of three treatment groups and asked
to indicate which, if either, of the two baskets of tomatoes they are most likely to
purchase, as well as how much they are willing to pay for both kinds of tomatoes. Using
this information, differences in average WTP based on COOL scenarios can be
estimated.
After participants indicate which labeled tomatoes they prefer to purchase and
how much they are willing to pay, they are asked to identify the reasons why they
selected a tomato (if they preferred one). This is first asked in an unaided format. After
answering these questions, participants are asked whether they noticed the different
origins of the tomatoes, what kind of information on the produce label they think is
important, and their general consumption preference toward tomatoes from different
production origins.
Demographics questions are answered by participants in the end of the survey.
After the participants complete the survey, the staff members who observed the
participants complete the survey will answer several questions about whether and how
the participant touched the tomatoes.
As both Florida and Mexico tomatoes are being used for this experiment, it is
important to collect data in multiple locations (in this case, Florida, Texas, and
Maryland). It is expected that WTP for Florida/US tomatoes compared to Mexico
tomatoes will be highest in Florida. Texas is selected because it is very close to Mexico
and participants are likely to see Mexico tomatoes more frequently and be familiar with
57
them. Maryland is selected as a region that does not have a reason to have a focus on
either Florida or Mexico, and thus serves as a type of control in this study. Specifically,
the cities selected are Tampa (Florida), Dallas (Texas), and Baltimore (Maryland). A
total of 210 intercepts are collected in each location (70 of each label treatment).
The contingent valuation method (CVM) is used in this survey to estimate
consumer WTP for fresh tomatoes. CVM is one of the most important valuation
methods to elicit the market value of new-products or non-market resources such as
environmental goods or services (e.g., Carson et al., 1992; Hanemann, 1994). CVM is
applied in a wide range of fields, including environmental protection, health care, and
food products (Diener et al., 1998; Hudson & Hite, 2003). Using a simulated or
hypothetical market to evaluate consumer preferences via directly asking their WTP for
certain goods or services in a survey, the CVM is called “contingent” because
consumers are asked to indicate their WTPs contingent on specific hypothetical market
scenarios.
Surveys based on CVM consist of three basic parts (Carson et al., 1989). First, a
hypothetical market under which the goods or services are to be offered is described
and presented to the survey participants. Then, questions about respondents’ WTP for
certain goods or services are presented and valued. Finally, socio-economic,
demographic characteristics questions about the consumers and their attitudes or
awareness toward the goods or services are answered by the respondents.
Haab and McConnell (2002) concluded that there are several inquiry approaches
to asking preference and WTP in CVM surveys, including open-ended format, bidding
games, payment cards, and dichotomous or discrete choice CVM. In this study, the
58
open-ended format is applied in the CVM survey analysis. In this case, the respondent
is asked directly to provide a point estimate of his or her WTP for certain goods or
services. One problem of the open-ended CVM is that the consumers encounter
difficulty stating their own price. Munro and Sugden (2003) indicated that consumer
preferences were dependent on reference prices. Monroe (1977) indicated that
consumers refer to reference price points to shape their own valuation of a product.
Chernev (2003) found that the articulation of reference prices before the choice can
simplify consumer preferences through imposing a structure consistent with the nature
of the decision task.
In this study, consumers are first asked about their purchasing choice for two
different labeled tomatoes in one of the three scenarios, then they are asked to name
their own WTP for the presented tomatoes given a range of reference prices. For
example, in scenario 1 the choice is between tomato #599 and tomato #280:
“Please go to the two baskets of tomatoes and consider them as if you were
deciding what to purchase. Afterwards, please answer the following questions about the
tomatoes:
Assume that you are going to the store to purchase one (1) pound of fresh
tomatoes. Which tomato do you prefer to purchase (if they were the same price)?
⃝Tomato #599 ⃝Tomato #280
If you needed to buy tomatoes and saw these, how much would you be willing to
pay per pound of tomatoes? Prices for tomatoes like these are usually $0.99–$3.99 per
pound. If you are not willing to purchase either or both, you can enter $0.00.
Tomato #599: $_____/lb Tomato #280: $_____/lb”
59
The reference price range refers to national, regular fresh tomato retail prices
from January 4, 2013 to March 14, 2014, acquired from the Agricultural Marketing
Service of the United States Department of Agriculture (USDA/AMS, 2013).
Logistic models are used to determine the relationship between consumer choice
of tomatoes and demographics, as well as various tomato COOL scenarios. The impact
of COOL scenarios on consumer choice of tomatoes can be determined by testing the
significance of the coefficients of the COOL scenario variables. The interaction between
production origin and COOL scenarios is also included in the model to determine
whether production origin information has different impacts on consumer choice under
different scenarios. Consumer’s WTP for Florida/US tomatoes versus Mexico tomatoes
is estimated. The reasons that consumers choose particular tomatoes and the impact of
production origin information on their choice are documented and compared across
different label scenarios.
Descriptive Analysis of the Survey Data
Demographic of Participants
A summary of the demographic characteristics of the participants is presented in
Table 4-1. After screening for respondents who did not qualify as an adult (18+ years
old), primary grocery shopper who purchased fresh tomatoes at least once/month in the
past few months, a total of 632 respondents completed the survey, including 209, 210,
and 213 samples in Baltimore, Dallas, and Tampa, respectively. Females and males
accounted for 55.5% and 44.5% of the total respondents, respectively. Most participants
in the sample were less than 40 years old (average, 36 years old). As for ethnicity,
Caucasians accounted for 51.7%, followed by Black or African-American (34.2%),
Hispanic (16.0%), and other races (6.1%). People with some college or a four-year
60
college degree were the largest proportion of the respondents at 53.0%, followed by
people with a high school diploma or GED (33.1%). The largest group of the participants
had full-time jobs (46.9%), followed by part-time jobs (18.7%). Counting members of the
household, 50.8% of the participants had 2–3 people in their household, 31.0% had 4–6
people in their household, and 14.9% lived alone. About 45.9% of the participants had
at least one child in the family; of those households with children, 24.1% had two or
more children and 21.8% had only one child. Those participants who refused to indicate
their annual household income accounted for about 14.9% of the total participants,
while the average estimated household income of participants who answered the
question was in the $50,000–$74,999 range. The results also show that 27.9% of the
respondents averaged spending $100–$149 per week on food at the grocery store,
21.4% spent $50–$99, and 19.6% spent $150–$199; the average food cost at grocery
stores in the survey was in the $150–$199 per week range.
Consumers’ Purchasing Habits of Fresh Tomatoes
In the survey, consumers were required to answer basic questions about their
purchasing habits and attitudes toward fresh tomatoes. The survey results show that
45.4% of the respondents indicated that they had bought fresh tomatoes once per week
within the past few months. Approximately 20.9% and 18.0% indicated that they
purchased fresh tomatoes 2–3 times per month and more than once per week,
respectively. As for the location where they usually purchased fresh tomatoes, 64.7% of
the respondents indicated that they buy from supermarkets, 51.5% from local grocery
stores and 24.5% from farmer’s markets. Another 13.1% indicated that they purchase
fresh tomatoes from a warehouse or roadside stand. Respondents identified regular
61
tomatoes (42.3%) and tomatoes on the vine (18.7%) as the most frequently purchased
types of tomatoes. Other tomato choices included heirloom, grape, Roma, and cherry.
When asked to identify what factors are most important when purchasing
tomatoes, respondents indicated freshness, firmness, and color as the top three factors.
Price, tomato size, and shape were relatively less important, and variety, country of
origin, on the vine or not, and availability of samples were the least important factors.
Consumers’ Attitudes and Preference of Different Labeled Fresh Tomatoes
After being given the opportunity to look at and touch the tomatoes in the
experiment, respondents were asked about their choice and attitude toward different
labeled tomatoes (Table 4-2). In scenario one, 56.8% of the respondents chose the
tomato with the US sticker, 24.2% chose the tomato with the Mexico sticker, and 19.0%
indicated no preference. In scenario two, 57.6% of the respondents chose the tomato
with the Florida sticker, 30.5% chose the tomato with the Mexico sticker, and 11.9%
indicated no preference between the two kinds of tomatoes. In scenario three, 59.7% of
the respondents chose tomato with the “Grown in Florida” sign on top of the basket,
29.4% chose the tomato with the Mexico sticker, and 10.9% indicated no preference.
In total, 44.3% of the respondents indicated that they had noticed the stickers or
signs and 55.7% had not. Specifically, 35.6%, 42.9%, and 54.5% of respondents in
scenarios one, two, and three, respectively, noticed the stickers or sign.
Participants were asked about what kinds of information they typically look for
when they purchase fresh produce. Nearly one-third indicated that they generally do not
look at labels on fresh produce. For the consumers who usually looked at the labels,
they focused on organic information (48.4%), brand (46.5%), country of origin (43.7%),
and nutrition information (33.4%).
62
Finally, participants were asked directly if they preferred tomatoes grown in the
United States to ones grown in Mexico. In this case, 48.0% indicated that they preferred
US tomatoes over Mexico tomatoes. Similarly, 49.2% preferred tomatoes produced in
Florida compared to tomatoes from Mexico. When asked about preferences between
tomatoes grown in Florida or the United States, more than half (56.8%) had no
preference. This information differed by location, with 64.8% of respondents in Tampa
preferring tomatoes produced in Florida over tomatoes produced in Mexico. This
compares to respondents in Baltimore (48.3%) and Dallas (34.3%). This also occurred
with tomatoes produced in Florida compared to tomatoes produced in the United States,
with 43.7% of respondents in Tampa preferring Florida-grown tomatoes compared to
23.9% in Baltimore and 17.1% in Dallas.
Consumers’ WTP for Florida/US Tomatoes and Mexico Tomatoes
Immediately following looking at the tomatoes and indicating which they preferred
(if either), participants were asked to indicate what they would be willing to pay for each
tomato they saw (Table 4-3). In scenario 1, participants were willing to pay an average
of $1.87/lb for the tomato with the US sticker and $1.55/lb for the tomato with the
Mexico sticker. In scenario 2, participants were willing to pay an average of $1.81/lb for
the tomato with the Florida sticker and $1.63/lb for the tomato with the Mexico sticker. In
scenario 3, participants were willing to pay an average of $1.68/lb for the tomato with
“Grown in Florida” sign plus US sticker and $1.50/lb for the tomato with the Mexico
sticker.
Models
Latent preference plays an important role in consumer purchasing behavior.
Specifically, in this experiment, latent preference includes consumer purchasing habits,
63
general attitudes of consumers toward different attributes of fresh tomatoes in their true
purchasing behavior, and the perception of COOL for fresh tomatoes. Three different
COOL scenarios, demographic characteristics, and latent preferences of the consumers
are assumed to affect consumer behavior regarding COOL observation/interest in the
experiment. Furthermore, whether or not the consumers read the tomato COOL as
another determining factor other than within the COOL scenarios, latent preference, and
demographics is also assumed to have an impact on consumers’ final choice and WTP
for Florida/US tomatoes vs Mexico tomatoes.
First, a binary logistic regression model is performed to determine the elements
which affect consumer behavior related to reading COOL information on the tomatoes.
Then an ordered logistic regression is used to determine the factors, including latent
preference, demographics, and reading COOL information, that influence consumer
choice of the different labeled tomatoes. Finally, a linear regression is estimated with
ordinary least square methods (OLS) to determine factors that impact consumers’ WTP
for Florida/US and Mexico tomatoes. In the binary logistic, ordered logistic and OLS
regression models, effective variables are screened and selected through stepwise
processes. A significance level of 0.50 is required to allow a variable into the model, and
a significance level of 0.15 is required for a variable to stay in the model.
With all effective variables detected through stepwise processes, simultaneous
equations are conducted between the behavior of reading COOL and consumers’
choice as well as the difference in WTP. Through running the simultaneous equations,
the endogenous variable Orginread can be controlled so the estimated results are more
valid and reasonable.
64
Theoretically, logistic regression is generally used to analyze the relationship
between a discrete response and a set of explanatory variables. For a binary logistic
model, the outcome variable 𝑌 is with has two possible categorical outcomes, denoted
by 1 and 0 (e.g., 𝑌 =1 if an event occurred, otherwise 𝑌 =0). Now suppose 𝑿 is a vector
of explanatory variables and the probability of 𝑌 =1 is defined as 𝜋 = 𝑃𝑟(𝑌 = 1|𝑋), then
the binary logistic regression has the form
𝐿𝑜𝑔𝑖𝑡(𝜋) = log (𝜋
1 − 𝜋) = 𝛼 + 𝜷𝑿
where 𝛼 is the intercept and 𝜷 is the vector of the slope parameters of the explanatory
variables.
The ordered logistic model is a regression model for ordinal outcomes. In an
ordered logistic model, there is an observed ordinal variable 𝑌𝑖, which, in turn, is a
function of another variable 𝑌∗ which is not measured. 𝑌∗ is a continuous, unmeasured
latent variable and its values determine the corresponding observed ordinal variable 𝑌.
Since 𝑌∗ cannot be measured, only its categories (from 1 to 𝑁) of outcomes are
observed:
𝑌𝑖 =
{
1 𝑖𝑓 𝑌𝑖
∗ ≤ 𝜇1 ,
2 𝑖𝑓 𝜇1 < 𝑌𝑖∗ ≤ 𝜇2,
3 𝑖𝑓 𝜇2 < 𝑌𝑖∗ ≤ 𝜇3,
… 𝑁 𝑖𝑓 𝑌𝑖
∗ ≥ 𝜇𝑁−1.
The ordered logistic model can be written as:
𝑃(𝑌𝑖 > 𝑗) = g(𝐗𝜷𝒊) =exp (𝑎𝑗 + 𝑿𝒊𝜷𝒊)
1 + [exp (𝑎𝑗 + 𝑿𝒊𝜷𝒊)], 𝑗 = 1,2, … ,𝑁 − 1
where 𝑿 is a vector of explanatory variables and 𝑎𝑗 is the intercept and 𝜷 is the vector of
the slope parameters of the explanatory variables.
65
From the above, it can be inferred that
𝑃(𝑌𝑖 = 1) = 1 − g(𝑿𝒊𝜷𝟏)
𝑃(𝑌𝑖 = 𝑗) = g(𝑿𝒊𝜷𝒋−𝟏) − g(𝑿𝒊𝜷𝒋), 𝑗 = 2,… ,𝑁 − 1
𝑃(𝑌𝑖 = 𝑁) = g(𝑿𝒊𝜷𝑵−𝟏)
In this study, the empirical models are defined as follows: Model 1 is binary
logistic regression, Model 2 is ordered logistic regression, and Model 3 is linear OLS
regression:
Model 1: 𝑂𝑟𝑖𝑔𝑖𝑛𝑟𝑒𝑎𝑑 = 𝛼0 + 𝛼1𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 + 𝛼2𝐶𝑖𝑡𝑦 + 𝜶𝟑𝑿 + 𝜶𝟒𝑫𝒆𝒎𝒐 + 𝜀1
Model 2: 𝐶ℎ𝑜𝑖𝑐𝑒 = 𝛽0 + 𝛽1𝑂𝑟𝑖𝑔𝑖𝑛𝑟𝑒𝑎𝑑 + 𝛽2𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 + 𝛽3𝐶𝑖𝑡𝑦 + 𝜷𝟒𝑿 +
𝜷𝟓𝑫𝒆𝒎𝒐 + 𝜀2
Model 3: 𝑊𝑇𝑃𝐷 = 𝛾0 + 𝛾1𝑂𝑟𝑖𝑔𝑖𝑛𝑟𝑒𝑎𝑑 + 𝛾2𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜2 + 𝛾3𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜3 + 𝛾4𝐶𝑖𝑡𝑦2 +
𝛾5𝐶𝑖𝑡𝑦3 + 𝜸𝟔𝑿 + 𝜸𝟕𝑫𝒆𝒎𝒐 + 𝜀3
where 𝑿 is a vector of all possible latent preference variables and 𝑫𝒆𝒎𝒐 is the vector of
all demographic variables. Choice is an ordinal variable (3=choose Florida/US labeled
tomatoes, 2=no preference, and 1=choose Mexico tomatoes). WTPD is the willingness
to pay difference between Florida/US tomatoes and Mexico tomatoes. Dummy variables
are created for Scenario (1=tomato with the US sticker vs tomato with Mexico sticker,
2=tomato with Florida sticker vs tomato with Mexico sticker, and 3=tomato with Florida
sign plus US sticker vs tomato with Mexico sticker), City (1=Baltimore, 2=Dallas and
3=Tampa), Gender (1 for male and 2 for female), Employment (1=full-time, 2=part-time,
3=currently not working, 4=retired, 5=student, 6=unpaid family worker), and consumers’
general preference between tomatoes from different production origins when doing their
regular shopping (1=prefer US tomato, 2=prefer Mexico [MX] tomato and 3 is no
66
preference; 1=prefer FL tomato, 2=prefer MX tomato, and 3 is no preference; 1=prefer
FL tomato, 2=prefer tomato from other states, and 3=no preference).
After running the three models separately through stepwise processes, each
model has its respective vector of selected explanatory variables. Scenario, City,
Originread and are forced to be included as explanatory regardless of their significance.
𝒁𝟏, 𝒁𝟐, and 𝒁𝟑 are the vectors of significant independent variables for model 1, 2 and 3,
respectively. Two simultaneous equation systems are established as follows:
Model 4: {𝐶ℎ𝑜𝑖𝑐𝑒 = 𝛽′0 + 𝛽′1𝑂𝑟𝑖𝑔𝑖𝑛𝑟𝑒𝑎𝑑 + 𝛽′2𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 + 𝛽′3𝐶𝑖𝑡𝑦 + 𝜷′𝟒𝒁𝟐 + 𝜀′2𝑂𝑟𝑔𝑖𝑛𝑟𝑒𝑎𝑑 = 𝛼′0 + 𝛼′1𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 + 𝛼′2𝐶𝑖𝑡𝑦 + 𝜶′𝟑𝒁𝟏 + 𝜀′1
Model 5: {
𝑊𝑇𝑃𝐷 = 𝛾′0+ 𝛾′
1𝑂𝑟𝑖𝑔𝑖𝑛𝑟𝑒𝑎𝑑 + 𝛾′
2𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜2 + 𝛾′
3𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜3 + 𝛾′
4𝐶𝑖𝑡𝑦2
+𝛾′5𝐶𝑖𝑡𝑦3 + 𝜸′𝟔𝒁𝟑 + 𝜀′3𝑂𝑟𝑔𝑖𝑛𝑟𝑒𝑎𝑑 = 𝛼′0 + 𝛼′1𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 + 𝛼′2𝐶𝑖𝑡𝑦 + 𝜶′𝟑𝒁𝟏 + 𝜀′1
where dummy variables are also needed similar to models 1, 2, and 3. The results of
simultaneous equation systems are presented in detail in next part. SAS (Statistical
Analysis Software) is used to analyze the models presented above.
Results
Factors Influencing Whether or Not Consumers Read Tomato Cool
For the binary logistic regression results of Originread, the estimated parameters
are reported in Table 4-4. The results indicate that the consumer’s likelihood to read
COOL is affected by different COOL scenarios, consumers’ latent preferences, and their
demographic characteristics.
Scenario, as one of the independent variables, is statistically significant at the
level of 𝛼 = 0.10. Specifically, compared to the base scenario (scenario 1) in which the
format is US and Mexico stickers on tomatoes, the second scenario, Florida and Mexico
stickers on tomatoes, is positive but insignificant which means Florida stickers fail to
67
increase the probability of reading the COOL on the tomatoes when consumers
purchase fresh tomatoes. In the third labeling scenario, which includes the “Grown
Florida” sign plus US sticker and Mexico sticker, is positive and significant statistically
related to consumer’s behavior of reading COOL in the experiment.
Several variables about consumers’ latent preferences are identified and
estimated, and are statistically significant at the level of 𝛼 = 0.10. The results reflect that
the probability of reading the COOL will increase if consumers regard the production
origin as an important factor when they make a decision to purchase fresh tomatoes.
For consumers who are more concerned with visual attributes, such as tomato size,
color, the probability of reading COOL decreases. Firmness and Sample Availability are
also both positive and significant. It can be inferred that if firmness and sample
availability of fresh tomatoes are more important to consumers, then their possibility of
reading the tomato COOL increases.
Furthermore, it is also found that if consumers typically look for the information
about where the produce is produced when they purchase fresh produce (refers to
variable GLCOOL), then they are more likely to read the COOL on the tomatoes in the
experiment; if they don’t read labels in their regular shopping for produce (refers to
variable GLdont), then the probability of reading tomato COOL will decline. Compared
to no preference, if consumers prefer to buy tomatoes from a certain producing area
(refer to variable preferUS, preferMX, preferOT), then they tend to read tomato COOL to
acquire production origin information.
For demographic variables, City, Age, EBlack, EAian, Education, Children18, are
statistically significant at the level of 𝛼 = 0.10. The estimated parameter of City3,
68
referring to Tampa, is positive and significant compared with the base city, Baltimore;
City2, Dallas, is positive but insignificant. It can be concluded that consumers living in
Tampa have a higher probability of reading tomato COOL than have consumers in
Dallas and Baltimore. The negative and significant coefficient for Age implies that the
older the consumer is, the less likely he/she reads the COOL. Consumers with higher
education also have a higher probability of reading the COOL. Black or African-
American consumers tend to read tomato COOL more often than do American Indian or
Alaska-Native consumers. Consumers who have more children under 18-year-old are
more likely to read tomato COOL.
Factors Affecting Consumer’s Choice for Tomatoes with Different Labels
The results (Table 4-5) of the first simultaneous equation system (model 4)
present the factors that have statistically significant influences on consumer’s
purchasing choice of fresh tomatoes. The dependent variable Choice has ordered
values that “3” refers to tomatoes labeled with the US sticker, the Florida sticker, or the
“Grown in Florida” sign plus US sticker; “2” refers to no preference; and “1” stands for
tomatoes labeled with Mexico sticker. Somewhat surprisingly, both the scenario and city
variables are all insignificant at the level of 𝛼 = 0.10. This implies that different labeling
strategies and where the participant lives have no impact on consumer’s choice for
fresh tomatoes. However, the variable Originread is positive and significant which
reveals that reading the tomato COOL practically increases the probability for
consumers to choose Florida/US tomatoes no matter what labeling strategy is applied. If
controlling for the other explanatory variables, consumers who read the tomato COOL
69
have 163.7% (exp(�̂�1) = 1.6366) higher odds than do those who do not of choosing
Florida/US tomatoes.
Color is one of the most important tomato attributes, with a positive and
statistically significant estimated parameter. It implies that consumers who are more
concerned with color when they purchase tomatoes have a higher probability of
choosing Florida/US tomatoes.
PreferMX (i.e., consumers who prefer purchasing Mexico tomatoes when they do
regular shopping) has a negative and significant coefficient. This can be interpreted as
meaning that those who usually prefer Mexico tomatoes have a higher probability of
choosing Mexico tomatoes instead of Florida or US tomatoes. PreferFLL is positive and
significant, which also follows, as it implies that consumers who usually prefer tomatoes
produced in Florida rather than in the United States in general are more likely to choose
Florida/US tomatoes.
For demographic variables, Caucasian consumers tend to choose Florida/US
tomatoes. Education also affects consumers’ choice that those with higher levels of
education are relatively less likely to choose Florida/US tomatoes.
Whether or not the participant touched the tomatoes when determining which to
purchase may play a key role in consumer choice. Touching tomatoes provides
consumers with further information about their attributes like firmness and size, and it
might increase the possibility for consumers to notice COOL on the tomatoes. Results
indicate that if they touched one of the tomatoes labeled as US or Florida, they were
more likely to choose Florida/US tomatoes, while if they touched the tomatoes labeled
from Mexico, they were more likely to choose Mexico tomatoes.
70
Factors Affecting Consumer’s WTP: A Premium for Florida/US Tomatoes
The difference between WTPs for Florida/US tomatoes (#599, #462, and #828)
and Mexico tomatoes (#280) is the premium that consumers are willing to pay for
Florida/US tomatoes, namely the dependent variables WTPD. If WTPD>0, it means
people are willing to pay more to purchase Florida/US tomatoes; if WTPD=0,
consumers are unwilling to pay more for either tomato; and if WTPD<0, people are
willing to pay more on Mexico tomatoes compared to Florida/US tomatoes. The results
of the regression on WTPD under the second simultaneous equation system (model 5)
are displayed in Table 4-6.
Surprisingly, for the Scenario variable, it is found that compared with scenario
one, the other two COOL scenarios have negative and significant estimated
parameters. The result indicates that applications of the Florida sticker and “Grown in
Florida” sign plus US sticker labeling strategies actually decrease the premium
consumers are willing to pay for Florida/US tomatoes compared to Mexico tomatoes
when compared to the base strategy of a US sticker.
Compared with the base city, Baltimore, residents in Dallas and Tampa are not
likely to behave differently with respect to WTPD. This result implies that there is no
difference in WTPD between residents of different cities.
The variable Originread was found to be statistically insignificant related to
WTPD, indicating those who read COOL were no more likely to pay more or less than
they would for Florida/US tomatoes.
The location where consumers generally purchase fresh tomatoes does impact
WTPD. People who usually purchase fresh tomatoes from a local store or farmers’
market are willing to pay a lower premium for Florida/US tomatoes compared to those
71
who purchase from a warehouse. Another location variable, grocery store, has no
impact on WTPD.
Search attributes for fresh tomatoes, including size, freshness, and on vine or
not, affect the premium that consumer are willing to pay for Florida/US tomatoes. As the
importance of tomato size and whether tomatoes are on the vine or not increases, the
premium for Florida/US tomatoes increases. The importance of tomato freshness has a
negative effect on WTPD, indicating that consumers who feel freshness is very
important have a smaller difference between WTP for Florida/US and Mexico tomatoes.
The coefficient for PreferMX is negative and statistically significant. It is
reasonable because people who usually prefer Mexico tomatoes are willing to pay a
smaller difference between the two types of tomatoes.
Females are generally willing to pay a higher premium for Florida/US tomatoes
than are males. The results also show that the more children under the age of 18 years
old a family has, the smaller difference between Florida/US tomatoes and Mexico
tomatoes the family is willing to pay.
Whether or not the consumer touched the tomatoes does influence WTPD. If
consumers touched Florida/US tomatoes in the experiment, they were willing to pay a
higher premium for Florida/US tomatoes; if they touched tomatoes from Mexico, the
premium for Florida/US tomatoes decreased.
72
Table 4-1. Sample demographic descriptive statistics
Variable Variable Description
Sample % (N=632)
US Census 2012
Gender Male 44.5 49.0 Age 18–30 50.3 13.9
31–50 30.0 26.5 51–70 15.0 23.9 70+ 4.3 9.1
Race Caucasian 51.7 63.0 Black or African American 34.2 13.1 Hispanic 16.0 16.9 Native Hawaiian or Pacific Islander 1.1 0.2 Asian 2.5 5.1 American Indian or Alaska Native 3.0 1.2
Income Less than $14,999 10.6 13.0 $15,000–$24,999 14.7 11.7 $25,000–$34,999 15.8 10.7 $35,000–$49,999 15.7 13.6 $50,000–$74,999 15.0 17.5 $75,000–$99,999 5.7 11.7 $100,000–$149,999 4.8 12.5 $150,000–$199,999 1.0 5.0 $200,000 or above 1.9 4.5 prefer not to answer 14.9 N/A
Education Less than high school 4.1 12.6 High school degree or equivalent 33.1 30.0 Some college 36.2 29.0 Four-year college degree 16.8 18.7 Postgraduate 6.0 10.0 Trade/technical school 2.4 N/A
Employment Full time 46.8 47.2 Part time 18.7 11.4 Currently not working 13.9 5.1 Retired 10.9 N/A Student 7.6 N/A Unpaid family worker 2.1 0.01
House people 1 14.9 27.4 2–3 50.8 50.0 4–6 31.0 21.5 7–9 2.7 N/A 9 or above 0.6 N/A
Children None 54.1 67.7 under 18 1 21.8 13.8
2 14.9 11.7 3 5.5 4.7
73
Table 4-1. continued
Variable Variable Description
Sample % (N=632)
US Census 2012
4 or more 3.6 2.1 Money on food
per week Less than $49 5.9 N/A
$50–$99 21.4 N/A $100–$149 27.9 N/A $150–$199 19.6 N/A $200–$249 11.6 N/A $250–$299 6.7 N/A $300–$349 3.3 N/A $350–$399 1.1 N/A $400–$449 0.5 N/A $450–$499 0.6 N/A Above $500 1.6 N/A
Table 4-2. Consumers’ stated choice of different labeled tomatoes, sorted by scenario and by city
By scenario By city
1 2 3 Baltimore Dallas Tampa
Florida/US tomatoes 56.9% 57.6% 59.7% 59.8% 58.4% 55.9% No preference 19.0% 11.9% 10.9% 14.4% 8.6% 18.8%
Mexico tomatoes
24.2% 30.5% 29.4% 25.8% 32.9% 25.4%
Sample Size 211 210 211 209 210 213
Table 4-3. Willingness to pay for Florida/US and Mexico tomatoes, sorted by scenario and by city (Unit: $/lb)
Scenario 1 Scenario 2 Scenario 3
Tomato with the
US sticker
Tomato with the Mexico sticker
Tomato with the Florida sticker
Tomato with the Mexico sticker
Tomato with the
Florida sign plus US sticker
Tomato with the Mexico sticker
All city 1.88 1.55 1.81 1.63 1.68 1.50
Baltimore 2.05 1.68 1.81 1.56 1.77 1.47 Dallas 1.66 1.42 1.66 1.47 1.75 1.69 Tampa 1.96 1.56 1.94 1.85 1.50 1.33
74
Table 4-4. Parameter results of the binary logistic regression of consumers’ behavior of reading tomato COOL information in the experiment
Variable Parameter
Estimate
Standard
Error
Pr > ChiSq
Intercept –1.5211 0.7087 0.0318** SC2 0.3465 0.2341 0.1388 SC3 1.0092 0.2338 <.0001*** City2 0.0176 0.2413 0.9419 City3 0.8486 0.2652 0.0014*** Size –0.1787 0.0865 0.0388**
Firmness 0.2177 0.1302 0.0944* Color –0.2901 0.1276 0.0230** COOL 0.2081 0.0753 0.0057***
SampleA 0.1375 0.0711 0.0531* GLCOOL 0.8991 0.2395 0.0002*** GLdont –0.8226 0.2378 0.0005***
preferUS 0.4281 0.2373 0.0713* preferMX 0.8310 0.4504 0.0650* preferFLL –0.2616 0.2596 0.3135 preferOT 0.5736 0.3031 0.0584*
Age –0.0248 0.00697 0.0004*** EBlack 0.3624 0.2089 0.0828* Eaian –1.2050 0.6062 0.0468**
Education 0.1558 0.0812 0.0548* Children18 0.2277 0.0896 0.0111** Touch280 0.3276 0.1965 0.0954*
Number of Observations Used: n=632. Adj R-Sq: 0.3093 “*”: Significant at 10%; “**”: Significant at 5%; “***”: Significant at 1%
75
Table 4-5. Parameter results of the ordered logistic regression of consumers’ stated choice of tomatoes from the first simultaneous equation system
Variable Parameter Estimate
Standard Error
t Value Pr > |t|
Intercept –0.007825 0.303387 –0.03 0.9794 SC2 –0.087207 0.122097 –0.71 0.4751 SC3 –0.158146 0.130296 –1.21 0.2248 city2 –0.068993 0.126151 –0.55 0.5844 city3 –0.195120 0.126640 –1.54 0.1234
Originread 0.492628 0.244964 2.01 0.0443** Color 0.137895 0.055295 2.49 0.0126**
preferUS –0.078023 0.128182 –0.61 0.5427 preferMX –0.667956 0.244071 –2.74 0.0062*** preferFLL 0.230986 0.136483 1.69 0.0906* preferOT –0.067525 0.163508 –0.41 0.6796
ECaucasian 0.263867 0.105357 2.50 0.0123** Education –0.079685 0.042531 –1.87 0.0610* Touch280 –0.346921 0.152453 –2.28 0.0229**
TouchUSFL 0.503246 0.151024 3.33 0.0009*** _Limit2.Choice 0.398701 0.039872 10.00 <.0001
_Rho –0.135687 0.160307 –0.85 0.3973
Number of Observations Used: n=632 “*”: Significant at 10%; “**”: Significant at 5%; “***”: Significant at 1%
76
Table 4-6. Parameter results of the linear regression of consumers’ willingness to pay for a premium for Florida/US tomatoes than for Mexico tomatoes from the second simultaneous equation system
Variable Parameter Estimate
Standard Error
t Value Pr > |t|
Intercept 0.526567 0.296977 1.77 0.0762* SC2 –0.206582 0.107654 –1.92 0.0550* SC3 –0.215765 0.112239 –1.92 0.0546* city2 –0.066247 0.109195 –0.61 0.5441 city3 –0.051987 0.108406 –0.48 0.6315
Originread 0.180450 0.202484 0.89 0.3728 Llocalstore –0.147851 0.086548 –1.71 0.0876*
Lwarehouse 0.342576 0.191369 1.79 0.0734* Lfarmermarket –0.173357 0.102640 –1.69 0.0912*
Size 0.084085 0.039457 2.13 0.0331** Freshness –0.093470 0.054578 –1.71 0.0868* Vineornot 0.057727 0.030966 1.86 0.0623* preferMX –0.352666 0.196976 –1.79 0.0734* Female 0.154841 0.088115 1.76 0.0789*
Children18 –0.097164 0.039769 –2.44 0.0146** Touch280 –0.328790 0.130942 –2.51 0.0120**
TouchUSFL _Sigma.WTPD
_Rho
0.240007 1.060916
–0.004567
0.127766 0.030130 0.125099
1.88 35.21 –0.04
0.0603* <.0001 0.9709
Number of Observations Used: n=632 “*”: Significant at 10%; “**”: Significant at 5%; “***”: Significant at 1%
77
CHAPTER 5 CONCLUSIONS
Methyl bromide (MBr), as a soil fumigant, has been proven to be the optimal
fumigation strategy for its reliability, affordability, and effectiveness. Since it is being
phased out gradually under the Montreal Protocol, research has been conducted to
discover the technical efficacy of MBr fumigant alternatives under different
environmental condition, but the economic feasibility and reliability of research is limited.
The field trials done by the University of Florida (UF) have demonstrated the extent of
inconsistency of several alternative fumigation systems.
The partial budgeting results indicate that MBr:Pic (67:33) performs the best in
terms of economic feasibility and cost effectiveness compared to MBr:Pic (50:50) and
other alternative fumigation strategies, including TE-3, PicChlor 60, FL-3 way, and the
non-fumigated treatment. These results are based on data from scientific field trial
under strict technical conditions and assumptions. The calibrated usage of equipment
and material is guaranteed in the scientific trials. The results might be different for real
commercial tomato production. This is because cultural practices and actual costs and
returns vary from grower to grower due to different market situations, labor supply,
machine life and condition of equipment, managerial skill, and other factors (Sydorovych
et al., 2008). Referencing this study, tomato producers are advised to estimate their
own production, harvesting, and marketing costs based on individual farming and
financial situations so as to identify the optimal alternative MBr fumigation to be applied
to minimize potential financial losses.
This study also employs the stochastic dominance method to compare methyl
bromide and its alternative fumigation systems regarding the risk efficiency. The SSD
78
results show that the two MBr treatments dominate the other alternative in yield
performance; among the MBr alternatives, TE-3 dominates the rest of the alternatives.
The SERF result shows that after assuming initial wealth and farm size, MBr:Pic
(67:33), is the most risk efficient and preferred fumigation system within the given
relative risk-averse coefficient range, dominating other alternatives and non-fumigated
treatment. Among the MBr alternatives, TE-3 has a better performance in risk efficiency
than do PicChlor 60, the currently most popular fumigant in Florida tomato industry, and
the FL-3 way, which was developed by UF researchers.
In summary, our results suggest that current MBr alternative fumigation
strategies used in the Florida tomato industry fail to replace MBr fumigation treatments
effectively from the economics perspective. However, our trial is only short term and
based on scientific experiment. It is suggested that more research on MBr and its
alternatives should be conducted in the future to analyze the discrepancy in efficacy and
consistency, under various farm situations and over longer periods. Moreover, the
economic loss and added production costs due to repeated usage of MBr alternatives in
the Florida tomato industry also should be studied in future research.
This study also investigates the US consumer preference and willingness to pay
(WTP) for Florida/US tomatoes and Mexico tomatoes under different COOL scenarios.
The mall intercept survey results show that the majority (>55%) of the participants
chose Florida/US tomatoes rather than Mexico tomatoes or no preference. Additionally,
on average, consumers are willing to pay a higher premium for Florida/US tomatoes
than they are for Mexico tomatoes under all COOL scenarios.
79
As the characteristics in perceptions of country of origin labeling (COOL),
consumption pattern, purchasing behavior, and demographics become diverse, demand
and consumer WTP for Florida/US tomatoes and Mexico tomatoes will be significantly
different. The experiments results show that whether consumers read tomato COOL is
affected by different COOL strategies. The “Grown in Florida” sign plus US sticker can
increase the probability of reading COOL on the tomatoes when consumers purchase
fresh tomatoes. The survey also found that consumers living in Florida have a higher
probability of reading tomato COOL than do consumers in Dallas and Baltimore.
The effect of three market labeling stimuli regarding origin of production on
consumers’ demand and WTP for Florida/US and Mexico tomatoes is analyzed. Several
implications obtained from the results are important. The results suggest that different
COOL strategies have no significant impact on consumer’s choice of the tomatoes. It is
against the expectation that Florida sticker and “Grown in Florida” sign plus US sticker
labeling will increase the probability of choosing Florida/US tomatoes compared to US
sticker labeling on the tomatoes. The behavior of reading tomato COOL has a positive
and significant effect on the choice of tomatoes, meaning that if consumers notice the
COOL on the tomatoes, then they will be more likely to choose Florida/US tomatoes. It
is implied that the strategy of labeling with the “Grown in Florida” sign plus US sticker is
associated with consumer’ choice of tomatoes indirectly because it can first attract
consumers to read COOL on the tomatoes, thus the reading behavior improves the
probability of choosing Florida/US tomatoes. When it comes to WTP, the results further
indicate that the US sticker, Florida sticker, and “Grown in Florida” sign plus US sticker
strategies decrease the premium consumers are willing to pay for Florida/US tomato
80
compared to Mexico tomatoes. Under these two COOL scenarios, there is a smaller
difference between WTPs for Florida/US and Mexico tomatoes.
One of the limitations of this study is that because an open-ended CVM was
used in designing the survey, there may be a potential bias in estimating WTP.
Therefore, improved methods such as Choice Experiment or Experimental Auction may
be good choices for future studies.
81
LIST OF REFERENCES
Anderson, J. G., & Anderson, J. L. (1991). Seafood quality: Issues for consumer researchers. Journal of Consumer Affairs, 25(1), 144-163.
Anderson, J. R., & Dillon, J. L. (1992). Risk analysis in dry land farming systems (No. 2). Food & Agriculture Org.
Anderson, J. R., Dillon, J. L., & Hardaker, J. E. (1977). Agricultural decision analysis. Ames, IA: Iowa State University Press.
Babcock, B. A., Choi, E. K., & Feinerman, E. (1993). Risk and probability premiums for CARA utility functions. Journal of Agricultural & Resource Economics, 18(1).
Batra, R., Ramaswamy, V., Alden, D. L., Steenkamp, J. B. E., & Ramachander, S. (2000). Effects of brand local and nonlocal origin on consumer attitudes in developing countries. Journal of Consumer Psychology, 9(2), 83-95.
Becker, T., Benner, E., & Glitsch, K. (2000). Consumer perception of fresh meat quality in Germany. British Food Journal, 102(3), 246-266.
Brown, C. (2003). Consumers' preferences for locally produced food: A study in southeast Missouri. American Journal of Alternative Agriculture, 18(4), 213-224.
Burton, M., Rigby, D., Young, T., & James, S. (2001). Consumer attitudes to genetically modified organisms in food in the UK. European Review of Agricultural Economics, 28(4), 479-498.
Byrd, M. M., Escalante, C. L., Fonsah, E. G., & Wetzstein, M. E. (2007). Feasible fumigant-herbicide system alternatives to methyl bromide for bell pepper producers. Journal of Agribusiness, 25(1), 31.
Campbell, B. L., Nelson, R. G., Ebel, R. C., Dozier, W. A., Adrian, J. L., & Hockema, B. R. (2004). Fruit quality characteristics that affect consumer preferences for Satsuma mandarins. HortScience, 39(7), 1664-1669.
Carpenter, J. E., Gianessi, L. P., & Lynch, L. (2000). The economic impact of the scheduled US phaseout of methyl bromide. National Center for Food and Agricultural Policy, Washington, D.C. Retrieved from http://www.ncfap.org/documents/mb.pdf
Carpio, C. E., & Isengildina‐Massa, O. (2009). Consumer willingness to pay for locally grown products: the case of South Carolina. Agribusiness, 25(3), 412-426.
Carson, R. T., & Mitchell, R. C. (1989). Using surveys to value public goods: The contingent valuation method. Resources for the Future, Washington DC.
82
Carson, R. T., Mitchell, R. C., Hanemann, W. M., Kopp, R. J., Presser, S., & Ruud, P. A. (1992). A contingent valuation study of lost passive use values resulting from the Exxon Valdez oil spill (No. 6984). University Library of Munich, Germany.
Chambers, G. M., Sullivan, E. A., Ishihara, O., Chapman, M. G., & Adamson, G. D. (2009). The economic impact of assisted reproductive technology: A review of selected developed countries. Fertility and Sterility, 91(6), 2281-2294.
Chernev, A. (2003). Reverse pricing and online price elicitation strategies in consumer choice. Journal of Consumer Psychology 13(1&2): 51-62.
Clemen, R. T. (1991). Making hard decisions. Boston, MA: PWS-Kent Publishing.
Curtis, K. R., Gumirakiza, J. D., & Bosworth, R. (2014). Consumer preferences and willingness to pay for multi-labeled produce at farmers’ markets. Journal of Food Distribution Research, 45(1), 14-20.
Darby, M. R., & Karni, E. (1973). Free competition and the optimal amount of fraud. Journal of Law and Economics, 16(1), 67-88.
Deepak, M. S., Spreen, T. H., & VanSickle, J. J. (1996). An analysis of the impact of a ban of methyl bromide on the US winter fresh vegetable market. Journal of Agricultural and Applied Economics, 28, 433-444.
Dentoni, D., Tonsor, G. T., Calantone, R. J., & Peterson, H. C. (2009). The direct and indirect effects of 'locally grown' on consumers' attitudes toward agri-food products. Agricultural & Resource Economics Review, 38(3), 384-396.
Diener, A., O'Brien, B., & Gafni, A. (1998). Health care contingent valuation studies: a review and classification of the literature. Health economics, 7(4), 313-326.
Duniway, J. M. (2002). Status of chemical alternatives to methyl bromide for pre-plant fumigation of soil. Phytopathology, 92(12), 1337-1343.
Eastwood, D. B., Brooker, J. R., & Orr, R. H. (1987). Consumer preferences for local versus out-of-state grown selected fresh produce: the case of Knoxville, Tennessee. Southern Journal of Agricultural Economics, 19(2), 183-194.
Ferrer, M. C. R., Fonsah, E. G., & Escalante, C. L. (2011). Risk-efficient fumigant-mulching system alternatives for bell pepper production. Journal of the ASFMRA, 2011, 162-171. Retrieved from http://purl.umn.edu/118963
Ferrer, M. C. R., Fonsah, E. G., Escalante, C. L., & Culpepper, S. (2010, February). Profitability efficiency analysis of methyl bromide fumigants and mulch systems alternatives for pepper production in Georgia. Southern Agricultural Economics Association Conference, Orlando, FL:
83
Fonsah, E., Yu, Y., Escalante, C., Culpepper, S., & Deng, X. (2010). Comparative yield efficiencies of methyl bromide substitute fumigants and mulching systems for pepper production in the Southeast. Journal of Agribusiness and Rural Development, 1(15), 55-65.
Gao, Z., Schroeder, T. C., & Yu, X. (2010). Consumer willingness to pay for cue attribute: the value beyond its own. Journal of International Food & Agribusiness Marketing, 22(1-2), 108-124.
Haab, T. C., & McConnell, K. E. (2002). Valuing environmental and natural resources: the econometrics of non-market valuation. Edward Elgar Publishing.
Hadar, J., & Russell, W. R. (1969). Rules for ordering uncertain prospects. American Economic Review, 59(1), 25-34.
Hanemann, W. M. (1994). Valuing the environment through contingent valuation. The Journal of Economic Perspectives, 19-43.
Hardaker, J. B., Richardson, J. W., Lien, G., & Schumann, K. D. (2004). Stochastic efficiency analysis with risk aversion bounds: a simplified approach. Australian Journal of Agricultural and Resource Economics, 48(2), 253-270.
Hudson, D., & Hite, D. (2003). Producer willingness to pay for precision application technology: Implications for government and the technology industry. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie, 51(1), 39-53.
Hueth, B. M., L. Lynch, B. McWilliams, M. Roberts, J. Siebert, D. Sunding, and D. Zilberman. (1997, November). Economic impact of banning methyl bromide use in California agriculture. International Research Conference on Methyl Bromide Alternatives and Emissions Reductions Conference, San Diego, CA. Retrieved from http://www.mbao.org/1997airc/030hueth.pdf
Kay, R. D., Edwards, W. M., & Duffy, P. A. (1994). Farm management. New York: McGraw-Hill.
Klein, J. G., Ettenson, R., & Morris, M. D. (1998). The animosity model of foreign product purchase: An empirical test in the People's Republic of China. Journal of Marketing, 62(1), 89-100.
Lancaster, K. J. (1966). A new approach to consumer theory. The Journal of Political Economy, 74(2), 132-157.
Lehman, J., Bacon, J. R., Toensmeyer, U., Pesek, J., & German, C. (1998). An analysis of consumer preferences for Delaware farmer direct markets. Journal of Food Distribution Research, 29, 84-90.
84
Lewis, K. E., & Grebitus, C. (2013). Consumers’ attitudes towards country of origin labeling for sugar. (No. 150162). Agricultural and Applied Economics Association Conference, Washington, D.C.
Loureiro, M. L., & Hine, S. (2002). Discovering niche markets: A comparison of consumer willingness to pay for local (Colorado grown), organic, and GMO-free products. Journal of Agricultural and Applied Economics, 34(3), 477-488.
Loureiro, M. L., & Umberger, W. J. (2003). Estimating consumer willingness to pay for country-of-origin labeling. Journal of American Agricultural and Resource Economics, 28(2), 287-301.
Loureiro, M. L., & Umberger, W. J. (2004, August). A choice experiment model for beef attributes: What consumer preferences tell us. American Agricultural Economics Association Conference, Denver, CO. Retrieved from http://purl.umn.edu/19931
Loureiro, M. L., & Umberger, W. J. (2007). A choice experiment model for beef: What US consumer responses tell us about relative preferences for food safety, country-of-origin labeling and traceability. Food Policy, 32(4), 496-514.
Lynch, L., & Carpenter, J. (1999). Economic impacts of banning methyl bromide: where do we need more research. Annual Research Conference on Methyl Bromide Alternatives and Emissions Reductions. Retrieved from http://mbao.org/mbrpro99.html
Lynch, L., & Carpenter, J. (1998). Economic impacts of switching from pre-plant fumigation with methyl bromide. Annual Research Conference on Methyl Bromide Alternatives and Emissions Reductions. Retrieved from http://www.mbao.org/1998airc/039lynch.pdf
Lynch, L., C. Greene, and C. Kramer-LeBlanc. (1997). Broadening support for 21st century IPM. Proceedings of the Third National IPM Symposium/Workshop (USDA/ERS), Washington, D.C.
Mabiso, A., Sterns, J., House, L., & Wysocki, A. (2005, July). Estimating consumers’ willingness-to-pay for country-of-origin labels in fresh apples and tomatoes: A double-hurdle probit analysis of American data using factor scores. In American Agricultural Economics Association Annual Meeting, Providence, Rhode Island (pp. 24-27).
McCluskey, J. J., & Loureiro, M. L. (2003). Consumer preferences and willingness to pay for food labeling: a discussion of empirical studies. Journal of Food Distribution Research, 34(3), 95-102.
Monroe, K. B. (1977). Objective and subjective contextual influences on price perception. In A.G. Woodside, J.N. Sheth, & P.D. Bennet (Eds.), Consumer and Industrial Buying Behavior, pp. 287-296. New York: North Holland Publishing.
85
Munro, A., & Sugden, R. (2003). On the theory of reference-dependent preferences. Journal of Economic Behavior & Organization, 50(4), 407-428.
NAPIAP (1993). The biologic and economic assessment of methyl bromide. National Pesticide Impact Assessment Program, United States Department of Agriculture, Washington, D.C.
NCFAP (2013). National Center for Food and Agricultural Policy website. http://www.ncfap.org/
Nebraska Tractor Test Laboratory. (2013). Tractor test reports. Retrieved from http://tractortestlab.unl.edu/c/document_library/get_file?uuid=0cb0188d-d0e5-4b31-97f8-f0b3ef91e285&groupId=4805395&.pdf
Nelson, P. (1970). Information and consumer behavior. The Journal of Political Economy, 78(2), 311-329.
Noling, J. W., Botts, D. A., & MacRae, A. W. (2013). Alternatives to methyl bromide soil fumigation for Florida vegetable production. EDIS #CV290. UF/IFAS Extension, Gainesville, FL.
Noling, J. W., Chandler, C. K., & Gilreath, J. P. (2005). Commercial field evaluations of methyl bromide alternatives in Florida strawberry. International Research Conference on Methyl Bromide Alternatives and Emissions Reductions. Retrieved from http://mbao.org/2005/05Proceedings/008NolingMBAO2005-1.pdf
Osteen, C. D. (2000). Economic implications of the methyl bromide phaseout (No. 33697). Economic Research Service, United States Department of Agriculture, Washington, D.C.
Roberts, W. S., & Swinton, S. M. (1996). Economic methods for comparing alternative crop production systems: A review of the literature. American Journal of Alternative Agriculture, 11(1), 10-17.
Rosskopf, E. N., Chellemi, D. O., Kokalis-Burelle, N., & Church, G. T. (2005, June). Alternatives to methyl bromide: A Florida perspective. APSnet Feature.
Rosskopf, E. N., Kokalis-Burelle, N., McSorley, R., & Skvarch, E. (2010). Optimizing alternative fumigant applications for ornamental production in Florida. EDIS #IN818. UF/IFAS Extension, Gainesville, FL.
Schooler, R. D. (1966). Product bias in the Central American Common Market. The International Executive, 8(2), 18-19.
Schumann, K. D., Richardson, J. W., Lien, G., & Hardaker, J. B. (2004, August). Stochastic efficiency analysis using multiple utility functions. American Agricultural Economics Association Conference, Denver, CO. Retrieved from http://purl.umn.edu/19957
86
Schupp, A., & Gillespie, J. (2001). Consumer attitudes toward potential country-of-origin labeling of fresh or frozen beef. Journal of Food Distribution Research, 32(3), 34-44.
Shimp, T. A., & Sharma, S. (1987). Consumer ethnocentrism: construction and validation of the CETSCALE. Journal of Marketing Research, XXIV, 280-289.
Spreen, T. H., Van Sickle, J. J., Moseley, A. E., Deepak, M. S., and Mathers, L. (1995). Use of methyl bromide and the economic impact of its proposed ban on the Florida fresh market fruit and vegetable industry. UF/IFAS Bulletin No. 898. University of Florida, Gainesville, FL
Stigler, G. J. (1961). The economics of information. The Journal of Political Economy, 69(3), 213-225.
Stoddard, C. S., LeStrange, M., Aegerta, B., Klonsky, K., & De Moura, R. (2007). Sample costs to produce fresh market tomatoes. University of California, Davis, CA. Retrieved from file:///G:/Dropbox/Thesis/Methyl%20Bromide/tomatofrmktsj07%20(3).pdf
Sydorovych, O., Safley, C. D., Welker, R. M., Ferguson, L. M., Monks, D. W., Jennings, K., ... & Louws, F. J. (2008). Economic evaluation of methyl bromide alternatives for the production of tomatoes in North Carolina. HortTechnology,18(4), 705-713.
Umberger, W. J., Feuz, D. M., Calkins, C. R., & Killinger‐Mann, K. (2002). US consumer preference and willingness‐to‐pay for domestic corn‐fed beef versus international grass‐fed beef measured through an experimental auction. Agribusiness, 18(4), 491-504.
UF/IATPC (2009). Fall Tomatoes: Estimated production costs in the Manatee/Ruskin area, 2008-2009. International Agricultural Trade and Policy Center, University of Florida, Gainesville, FL. Retrieved from http://www.fred.ifas.ufl.edu/iatpc/files/RuskinFallTomato09.pdf
UN/FAO (2013). FAOSTAT Database. United Nations, Food and Agriculture Organization, New York, NY. Retrieved from http://faostat3.fao.org/faostat-gateway/go/to/download/Q/QC/E
USBLS (2013). BLS website. United States Bureau of Statistics, Washington, D.C. Retrieved from http://www.bls.gov/ppi/
USDA/AMS (2013). Fresh fruit and vegetable shipments. United States Department of Agriculture, Agricultural Marketing Service, Washington, D.C. Retrieved from http://www.ams.usda.gov/AMSv1.0/getfile?dDocName=stelprdc5102593
USDA/ARMS (2013). ARMS Farm Financial and Crop Production Practices. United States Department of Agriculture, Agricultural Resource Management Survey, Washington, D.C. Retrieved from http://www.ers.usda.gov/data-products/arms-
87
farm-financial-and-crop-production-practices/tailored-reports-farm-structure-and-finance.aspx#Pf798820fcfff45fcb91212c39fff79fa_5_66iT0R0T0R0x7.
USDA/ERS (2013). U.S. Tomato statistics. United States Department of Agriculture, Economic Research Service, Washington, D.C. Retrieved from http://usda.mannlib.cornell.edu/MannUsda/viewDocumentInfo.do?documentID=1210
USDA/NASS (2013a). Vegetables annual summary. United States Department of Agriculture, National Agricultural Statistics Service, Washington, D.C. Retrieved from http://usda.mannlib.cornell.edu/usda/nass/VegeSumm//2010s/2014/ VegeSumm-03-27-2014.pdf
USDA/NASS (2013b). Farm labor. United States Department of Agriculture, National Agricultural Statistics Service, Agricultural Statistics Board, Washington, D.C. Retrieved from http://www.usda.gov/nass/PUBS/TODAYRPT/fmla1213.pdf
USDA/NASS (2013c). Census of Agriculture. United States Department of Agriculture, National Agricultural Statistics Service, Agricultural Statistics Board, Washington, D.C. Retrieved from http://www.agcensus.usda.gov/Publications/2012/Full_Report/Volume_1,_Chapter_1_State_Level/Florida/
VanSickle, J. J. (2008). Country of origin labeling for fruits and vegetables. Choices, 23(4), 43-45.
VanSickle, J.J., Brewster, C., & Spreen. T. H. (2000). Impact of a methyl bromide ban on the US vegetable industry. UF/IFAS Bulletin No. 333. University of Florida, Gainesville, FL
VanSickle, J. J., Evans, E. A., & Emerson, R. D. (2003). US-Canadian tomato wars: An economist tries to make sense out of recent antidumping suits. Journal of Agricultural and Applied Economics, 35(2), 283-296. Retrieved from http://purl.umn.edu/37962
Verlegh, P. W., & Steenkamp, J. B. E. (1999). A review and meta-analysis of country-of-origin research. Journal of Economic Psychology, 20(5), 521-546.
Warmann, G. W. (1995). Economics of liming Kansas cropland. KS/AES MF-2137. Kansas State University, Manhattan, KS. Retrieved from http://www.ksre.ksu.edu/bookstore/pubs/MF2137.pdf
Wirth, F. F., Love, L. A., & Palma, M. A. (2007). Purchasing shrimp for at-home consumption: The relative importance of credence versus physical product features. Aquaculture Economics & Management, 11(1), 17-37.
Wossink, G. A. A., & Osmond, D. L. (2002). Farm economics to support the design of cost-effective best management practice (BMP) programs to improve water
88
quality: Nitrogen control in the Neuse River Basin, North Carolina. Journal of Soil and Water Conservation, 57(4), 213-220.
Zhu, M., Guan, Z., & Wu, F. An overview of the US tomato industry. The Florida Tomato Institute Proceedings, 28 (Naples, FL).
89
BIOGRAPHICAL SKETCH
Xiang Cao was a graduate student of the Master of Science program in the Food
and Resource Economics Department at the University of Florida. Xiang was born in
Xining, Qinghai, China in 1991. He received his bachelor’s degree in international
economics and trade from the University of International Relations in China in 2012. He
received his master’s degree in food and resource economics from the University of
Florida in 2014. His major area of study is production economics and agricultural
marketing analysis. In fall of 2014, he will pursue a PhD degree in agricultural and
applied economics at Virginia Polytechnic Institute and State University (Virginia Tech)
in Blacksburg, VA.