International Commodity Trade, Transport
Costs, and Product Di↵erentiation
Colin A. Carter, James A. Chalfant, and Navin Yavapolkul⇤
June 23, 2015
Abstract
Transport costs and product quality have received increased attention in the international trade literature. Product quality is a
particularly important factor in international trade of high-valued
commodities. We observe that significant transport costs for a relatively high quality product represent a natural trade barrier. In this
case, transport costs may introduce product di↵erentiation, protecting
home-country production of the higher quality product and constraining the foreign exporter to shipping a lower quality substitute. With
di↵erential transport costs between a higher and a lower quality product, firms in both countries may gain by implicitly coordinating in an
increasingly segmented market and choosing a high-price strategy for
both products. Home-country producers clearly gain. Through product di↵erentiation in an oligopolistic market, the foreign producers
may also gain when the home producers exploit the transport cost advantage. The international orange juice market, with significant trade
volumes in both lower quality frozen-concentrated juice and higher
quality not-from-concentrate juice, provides compelling evidence supporting this model.
Key words: commodity trade, product di↵erentiaition, orange juice
⇤
Carter and Chalfant are Professors at Univ.
of California, Davis,
[email protected], [email protected]. Yavapolkul is Professor at Kasetsart
Univ. in Thailand, [email protected]. Carter is the corresponding author.
1
JEL codes: F12, L11, Q17
1
Introduction
The recent trade literature highlights the importance of both transportation
costs and product di↵erentiation (e.g., Hummels and Skiba (2004); Bauman
(2004); Feenstra (1988)). In addition, the connection between product quality and trade has been modeled (Feenstra and Romalis (2014); Crozet et al.
(1989); Hallak (2006)). This literature generally confirms the Alchian-Allen
theorem which states that, in an international trade context, importers’ purchases will switch toward relatively higher quality products, in response to a
rise in transportation costs. This occurs because the added per-unit transport cost decreases the relative price of the higher quality product. Such
an e↵ect has been confirmed in markets such as grapes (Alchian and Allen,
1967) and textiles (Irwin and Temin, 2001), and is often stated colloquially
as “shipping the good apples out” (Borcherding and Silberberg, 1978). In
these papers, it is typical to assume that transport costs remain relatively
constant between di↵erent levels of quality. However, Hummels and Skiba
(2004) mention that if transport costs rise faster than goods prices, a reverse
Alchian-Allen e↵ect is possible.
This paper proposes a model in which the high-quality good cannot be profitably shipped out to an export market with domestic production, due to a
2
natural trade barrier arising from very high transportation costs associated
with the high-quality good, compared to the low-quality good. For many
products, transport costs may well depend on the quality of the product.
Certainly, we expect that e↵ort to bring about reduced product damage and
the increased quality that results will cost more in shipping and handling,
so there may be a range of quality-transport cost combinations, reflecting a
positive relationship between transport cost and quality.
For certain goods, there are some levels of quality that are prohibitively expensive to ship, making the high-quality product uncompetitive in certain
markets. Indeed, this is the case with Brazilian orange juice exported to the
U.S. market. Orange juice can be shipped in lower-quality concentrated form
(i.e., with the water removed) or in higher quality, fresh juice form. Fresh
juice exports from Brazil are not competitive in the U.S. market (where there
is domestic juice production) due to the high cost of transport, compared to
concentrated juice. This provides an example demonstrating that di↵erential
transport costs thus protect higher quality products in the importing country from foreign competition. This means that the exporter ships out only
the lower quality products, which then compete in a di↵erentiated market
against higher quality goods in the importing country. The relatively high
cost of transporting high quality goods therefore leads to more di↵erentiated
consumption in the importing country, as in Echazu (2009).
Continuing to draw implications from the U.S. orange juice market, which
3
resembles a classic oligopolistic market characterized by product di↵erentiation, this paper shows that the presence of prohibitively high transport costs
will act as a commitment mechanism that ensures that the foreign producer
will export only the lower valued product, giving the high quality producer
in the importing country a certain degree of market power. Producers in
two countries will, in e↵ect, coordinate on an increasingly segmented market
in the home country, and each will choose a higher price strategy than they
would otherwise have followed. Thus, in essence, a natural barrier such as
transportation cost could facilitate tacit collusion.
The idea that a trade barrier may change the strategic relationship between
exporting and home-country firms originates with Eastman and Stykolt (1960).
Harris (1985) and Krishna (1989) have independently developed this idea in
the context of U.S. Voluntary Export Restraints on Japanese autos, showing
that the export quota imposed on Japanese cars facilitated higher prices for
both U.S. and Japanese firms. While domestic policy is commonly identified
as a distortion to competition in trade, a locational barrier in the form of
transport costs has rarely been treated in the literature as such. In this paper, transportation costs e↵ectively prevent the foreign firm from following
a home-country leader in engaging in product di↵erentiation and, therefore,
such costs can be viewed as a deterrent to competition.
It has been widely acknowledged that asymmetries in cost and production
characteristics can hinder collusion among firms (e.g. Scherer and Ross
4
(1990)). In particular, Gross and Holahan (2003) found that increases in
transportation costs tend to destabilize collusive agreements. However, whether
product di↵erentiation by quality hinders or facilitates collusion appears to be
ambiguous [Ross (1992), Martin (1996), and Raith (1996)]. This study introduces a new dimension to this literature, demonstrating that transportation
costs can introduce product di↵erentiation and a collusive outcome.
Before proceeding to the theoretical and simulation model in greater detail
and depth, the next section describes the international market for orange
juice and elaborates upon the motivation and the main findings of this paper.
2
The Orange Juice Market
World trade in orange juice is a classic oligopolistic market. The U.S. and
Brazil are the world’s largest producers of orange juice, accounting respectively for 30% and 60% of global production. Production in each country
is regionally concentrated, in the state of Sao Paulo in Brazil, and in the
state of Florida in the United States. Brazil is the world’s largest exporter
of orange juice, accounting for 85% of world exports. The U.S. consumes
over 40% of world orange juice production and is a net importer of orange
juice, purchasing between 20-35% of its consumption from abroad. Behind
the EU, the United States is the second largest market for Brazilian exports.
5
But unlike in the EU, a significant share of U.S. consumption is domestically
produced (Brown et al., 2009).
A small number of companies that operate in Brazil and Florida dominate
the orange juice market. These companies are highly vertically integrated,
since both size and scale represent an important competitive advantage in
the industry, particularly in bulk transportation of the juice. Table 1 demonstrates the fact that orange juice processing and trade are highly concentrated. The trade protection e↵orts by the Florida industry have been acknowledged to result in a highly concentrated industry [Braga and Silber
(1991); Hart (2004)].
The Herfindahl-Hirschman Index (HHI) indicates that the market shares of
processing firms in the U.S. and Brazil would be considered moderately concentrated and concentrated, respectively, according to the Horizontal Merger
Guidelines issued by the U.S. Department of Justice and the Federal Trade
Commission.1
[Table 1 about here.]
In recent years, the U.S. orange juice market has experienced a dramatic
shift in both consumption and production patterns, from the once-dominant
Frozen Concentrated Orange Juice (FCOJ) to Not From Concentrate (NFC)
fresh juice. Consumers generally view NFC as a higher quality product than
FCOJ. Despite selling for a premium, NFC accounted for 53 percent of consumption in 2011, rising from 34 percent in 1997. Meanwhile, the share
6
of NFC in total U.S. production of orange juice increased from 29 percent
in 1997 to 59 percent in 2011. Figure 1 reflects the dramatic increase in
the relative importance of NFC, with an apparent structural change around
2004-05.
We attribute the structural change in 2004-05 to a series of hurricanes that
hit Florida, reducing the size of the orange harvest from 242 million 90 lb.
boxes in 2003-04 to 129 million boxes in 2006-07, a 47% drop in the size of the
crop. Hurricane Charley hit Florida on August 13, 2004, followed by Frances
on September 5, 2004 and Jeanne on September 26, 2004. Then the orange
groves were struck again the next year when hurricane Wilma came through
in October 24, 2005. When the crop was shortened due to the hurricanes, the
processors continued to produce NFC at historical levels, and at the same
time lowered production of FCOJ. This shift in production shares is evident
from Figure 1.
In the U.S. market, when there was a production shift from FCOJ to NFC
by processing firms in Florida, transportation costs largely prevented Brazil’s
producers from making the same switch for sales into the U.S. market. Transportation costs are much higher for NFC than for FCOJ, due to the large
di↵erence in volume for fresh versus concentrated juice. In order to ship
equivalent volumes of NFC and FCOJ, six times the volume of NFC must be
shipped, because FCOJ has had the water removed (through evaporation),
while NFC has not.
7
The data in Table 2 support our story in a number of respects. First, Table
2 shows that Brazil has been able to ship a larger and larger share of NFC to
the EU, where the Brazilians are not competing against domestic production.
In 2004, 27% of the Brazilian juice exported to the EU was NFC and by 2012,
the share was over 62%. This suggests that Brazil has a certain degree of
market power in the EU, as it is the dominant supplier in that market and
chooses to export the higher quality product to Europe. However, at the
same time, Brazil fails to sell high quality NFC in the same proportions in
the U.S. market, where it cannot exercise the same degree of market power
because of U.S. domestic production that is now focused on NFC. Instead,
Brazil juice is primarily imported into the U.S. as the lower quality product,
FCOJ. In 2012, only 21.4% of Brazil’s exports to the U.S. were NFC.
[Figure 1 about here.]
Before NFC juice gained such a large share of the U.S. market, and despite
a relatively high import tari↵ [about 30 cents per gallon Single Strength
Equivalents (SSE) for FCOJ], Brazil competed e↵ectively in the U.S. market, due to its lower cost of production.2 Clearly, the U.S. orange juice tari↵
supports the price of FCOJ in the U.S., and a reduction in the tari↵ would
allow the Brazilian orange juice industry to capitalize on its lower production costs (Donovan and Krisso↵, 2004). However, the shift from FCOJ to
NFC production by domestic processors following the hurricanes in 20042005, along with consumers’ acceptance, provided Florida with additional
8
protection through a natural trade barrier. The shift may have resulted from
an implicit strategy adopted by the Florida citrus industry (brought on by
the hurricanes) to slow foreign competition in the presence of globalization
and the lowering of trade barriers. What seems to be less obvious is that
the increasingly segmented market may have been welcomed by Brazilian
producers as well.
[Table 2 about here.]
We believe that both countries’ processors experienced an increase in market
power, due to greater specialization and di↵erentiation of their products.
While adopting consumers presumably gain from greater availability of a
new product, welfare improvements are mitigated by the exercise of increased
market power. The precise welfare e↵ects will depend on the extent to which
consumers view NFC and FCOJ as imperfectly substitutable products, and
of course on the size of price increases.
In the sections that follow, we demonstrate that there is a potential gain to
Brazilian exporters associated with another country’s home product di↵erentiation strategy, even though Brazil cannot follow Florida by increasing
quality. Under an oligopolistic setting, product di↵erentiation in international markets can result in higher prices for both home and foreign firms.
Higher prices for both firms can be sustained, however, only if Brazil can
precommit to produce a less desirable product, FCOJ, instead of competing
with Florida in selling NFC into the U.S. market.
9
We begin the next section with a simple duopoly model in which firms simultaneously choose a product’s quality and its price. Initially, the scenario
where transport cost is zero is used as a base case and the equilibrium outcome is undi↵erentiated. We then move to a sequential price competition
duopoly model, arising from the introduction of a prohibitive transport cost
that justifies the development of a higher quality product (NFC) by one producer, while the other reacts by raising the price for its now lower-quality
product. Our theoretical section compares the results of the two models—a
simple duopoly versus a price leader duopoly where one firm has a bounded
level of quality. After characterizing the new equilibrium, Section 4 employs
a two-firm simulation to illustrate the results suggested by the theoretical
model. It is shown that the profits for both firms increase. Section 5 discusses implications and concludes.
3
A Two-Firm Model of Quality Choice
Our approach to modeling the orange juice market, and to characterizing
production decisions followed by Florida and Brazil, is to first construct a
simple duopoly model depicting product di↵erentiation. Florida and Brazil
are treated as two firms, for convenience, and assumed to compete strategically by choosing both price and quality. Initially, we treat the equilibrium
as the outcome of a duopoly producing a single product that is highly substitutable, namely FCOJ, and characterize the Nash equilibrium situation for
10
the two producers (countries).
Under this initial equilibrium, Florida would not be able to capture market
share by charging a higher price without some limitation on the quality chosen by Brazil. Without transport costs, Brazil could freely choose any price
and quality, undercutting Florida’s price for the same quality, or matching
or even leapfrogging Florida’s quality for the same price. Each firm’s price
and quality choice is optimal and Nash equilibrium ensures that neither will
deviate from the prevailing market outcome.
However, in reality Brazil’s quality choice is constrained by transport costs.
The equilibrium price and quality can now change, because Florida knows
that Brazil can strategically change its price, but at the same time it cannot
change its quality. The new equilibrium in which Florida makes investments
in infrastructure and marketing to expand sales of NFC relative to FCOJ
is then characterized by a duopoly equilibrium, as will be shown below, in
which Florida is a price leader and Brazil is a follower.
Due to the imperfectly competitive market that results from Brazil’s inability
to follow, the producers in Florida are able to gain from product di↵erentiation, shifting to ready-to-serve, NFC fresh juice. Assuming that consumers
have accepted NFC as higher quality, Florida’s advantage comes from increased quality that Brazil cannot duplicate because of prohibitive transport
costs.
We start with the characterization of the initial equilibrium, described for
11
simplicity in terms of one home firm and one foreign firm. Suppose that both
the foreign and home firms compete in delivering products that are close
substitutes, specifically FCOJ.3 Each firm faces a demand function Di (p, q)
for i = f, h, where p = (pf , ph ) is a price vector consisting of foreign price and
home price, with the usual assumptions that ensure
@D i
@pi
< 0 and
@D i
@pj
> 0 for
i 6= j = f, h. It is also assumed that the quality of a product can be quantified
into a single variable qi . The quality vector q = (qf , qh ) satisfies
@D i
@qj
@D i
@qi
> 0 and
< 0; i.e., an increase in own-product quality promotes demand, while an
increase in the other firm’s quality dampens (own) demand.
Note that consumer preference for these goods is postulated so that we deal
only with the “reduced form” of the demand function. Since consumers
generally view NFC as a close alternative to freshly squeezed juice, to a
large extent, NFC is arguably a higher quality product when compared to
FCOJ. Thus, our analysis here can be conveniently described as the vertical
product di↵erentiation problem pioneered by Gabszewicz and Thisse (1979)
and Shaked and Sutton (1982), rather than the horizontal di↵erentiation
model, which would be more suitable for other types of products, such as
di↵erentiated breakfast cereals or automobiles. In reality, there is not a
continuous range of quality choices, only a discrete choice by Florida not to
concentrate a larger share of its juice and therefore deciding to increase the
average quality of its production. Quality could, of course, vary around these
choices—convenience in packaging, vitamin- or calcium-fortification, etc.—
but we are interested in the discrete choice made by one country and not the
12
other, instead of the e↵ects of these other attributes.
Therefore, our demand specification is chosen for simplicity and convenience,
and our analysis does not depend crucially on assumptions from structuralform type models to gain additional insight. The reduced form demand
relates quantity demanded to prices and qualities and ignores the possibility
of firms choosing from a range of qualities. The relative position of qualities
chosen by firms is determined solely by Florida’s decision.
We will assume zero marginal cost of producing good i. The fixed cost
associated with increasing the quality of the product (i.e., di↵erentiation),
Ci (qi ), is assumed to be a convex function of product quality, where Ci 0 (qi ) >
0 and Ci 00 (qi ) > 0; it is increasingly costly to di↵erentiate, the greater the
di↵erentiation of the product.4
Thus, cost does not depend on level of output, and quality entails only a
fixed cost that is unrelated to quantity produced. Florida will elect to increase quality, as long as the increased revenues o↵set the cost of doing so.
For a detailed analysis of cost specification in product di↵erentiation models,
see Ronnen (1991), Motta (1993), and Lehmann-Grube (1997). We do not
explore cost considerations in detail. The specification of the cost of producing di↵erent levels of quality will of course determine the specific equilibrium,
in the di↵erentiated case, but we are considering only the option of increasing
quality through the adoption of NFC, not a range of quality improvements.
Ultimately, di↵erent specifications for the cost function do not alter the main
13
findings of this paper. It can be shown that all of the results below still hold
with more restrictive assumptions on market parameters.
We assume that firms first choose price and quality, and then quantities adjust to clear the market. Quality choice in our model is viewed as a short-run
variable, as referred to by Feenstra (1988), in the sense that firms simultaneously choose price and quality, as opposed to a sequential model in which
firms choose long-run quality in a first stage and then price in the second
stage, where the quality choice is sunk.5
A simultaneous model is a more realistic portrayal of many circumstances,
in which firms must decide on price at the same time that quality attributes
such as convenience, packaging, or perhaps labeling are chosen.
Thus, firm i maximizes profit with respect to pi and qi , taking pj and qj , as
fixed.6 The objective function of firm i is the following:
⇧i (pi , qi ; pj , qj ) = pi Di (p, q)
Ci (qi ).
(1)
The solution to this maximization problem is a function of the price and
quality chosen by the other firm; in other words, the firm’s reaction functions
are pi (pj , qj ) and qi (pj , qj ). They are implicitly defined by the following first
14
order conditions:
⇧ipi (pi , qi ; pj , qj ) = pi Dpi i + Di (p, q) = 0
⇧iqi (pi , qi ; pj , qj ) = pi Dqi i
Ci0 (qi ) = 0.
(2)
These two first order conditions state a Dorfman-Steiner optimal advertising
condition, if quality is replaced by advertising expenditure (Dorfman and
Steiner, 1954).7
For the remainder of the analysis, we will assume linear demand for analytical
tractability.8 With Ci000 = 0, both reaction functions pi (pj , qj ) and qi (pj , qj )
will also be linear and are uniquely defined if we assume |H⇧i | =
2Ci00 Dpi i
2
Dqi i > 0, i.e., strict concavity of the profit function.
With all the assumptions invoked, condition (2) implies that the reaction
functions in Figure 2 satisfy
@pi
> 0,
@pj
@qi
< 0,
@qj
@pi
<0
@qj
@qi
> 0.
@pj
(3)
See the Appendix for proofs.
These derivatives describe a set of strategies available to firm i for a given
price and quality chosen by firm j. As firm j raises its price, depending on
the degree of substitutability, firm i will want to increase its price to match.
15
Higher own price gives additional marginal benefit from increasing quality.
Firm i will adjust its quality, which in turn induces its own price to increase
further. On the other hand, if firm j were to raise its quality, this would take
away firm i’s market share. Firm i will therefore react by lowering its price
to attract more consumers. A smaller own price decreases marginal benefit
from di↵erentiating quality. Thus, firm i would gain more profit by reducing
quality. As a result, the price of firm i is a strategic complement and the
quality of firm i is a strategic substitute to those same strategic choices of
firm j.
[Figure 2 about here.]
The price reaction functions of foreign and home are depicted in the left
panel of Figure 2 as Pf Pf and Ph Ph . These reaction functions are drawn for
the given levels of quality qf and qh , denoted by pf (ph ; qh ) and ph (pf ; qf ),
respectively. Likewise, the quality reaction functions, denoted by qf (ph ; qh )
and qh (pf ; qf ), are shown in the right panel by Qf Qf and Qh Qh , respectively. Nash equilibrium in price and quality takes place when Ep (qf , qh ) and
E
NE
Eq (pf , ph ) are aligned at Ep (qfN E , qhN E ) and Eq (pN
f , ph ). More specifically,
16
the equilibrium is simply a solution of the following system of equations.
⇧fpf = pf Dpff + Df (p, q) = 0
⇧fqf = pf Dqff
Cf0 (qf ) = 0
⇧hph = ph Dphh + Dh (p, q) = 0
⇧hqh = ph Dqhh
Ch0 (qh ) = 0,
(4)
(5)
(6)
(7)
(Dpj i Ci00 + Dqji Dqi i ) for i 6= j = f, h to ensure
where we require |H⇧i | >
stable Nash equilibrium. The equilibrium solution of price and quality before
the shift to NFC will be determined by the structure of demand and cost
exhibited through the market parameters.
If the foreign firm wanted to di↵erentiate further and decided to ship NFC
instead of FCOJ, it would have to pay a very high transportation cost. Thus
we assume there is a constraint on the quality level that the foreign firm can
produce. Under reasonable assumptions, it may be impossible for foreign
producers to increase quality beyond a certain level, Q̄. That is,
@Cf
=
@qf
(
Cf0 (qf ) qf  Q̄f
1
(8)
qf > Q̄f .
[Figure 3 about here.]
With the quality restrictions in place, the foreign firm will instead maxi-
17
mize
⇧f (pf , qf ; ph , qh ) = pf Df (p, q)
Cf (qf )
subject to qf  Q̄f .
(9)
This yields the following first order conditions:
pf Dpff + Df (p, q) = 0
Cf0 (qf ) =
pf Dqff
(qf
where
(10)
(11)
Q̄f ) = 0,
0 and qf  Q̄f . Instead of assuming an arbitrary number, Q̄f
can be set at qfN E , which indicates that the foreign firm cannot move further
away in quality space from the Nash Equilibrium quality defined by (4) to
(7). Figure 3 illustrates the first-order condition (11) when
= 0 and qf =
E f
0
NE
qfN E = Q̄f . The two marginals are equal when pN
f Dqf = Cf (qf ). The
marginal cost of product quality for the foreign firm is depicted by the locus
oef . When
> 0 and qf = Q̄f , the foreign firm’s quality does not react
strategically to any change in qh . This is shown in Figure 2 along the line
Q̄f Eq , while along Eq Qf the quality constraint is not binding so that
=0
and qf < Q̄f . Figure 2 depicts the new quality reaction function, which has
been changed from Qf Qf to Q̄f Eq Qf .
18
In the case that
> 0, we substitute qf = Q̄f into (10) to obtain
pf Dpff + Df (p, Q̄f , qh ) = 0,
(12)
which implicitly defined p˜f (ph , qh ) as a function when qf is fixed. This new
function is represented by the locus P˜f P˜f in Figure 2. It can easily be shown
that p˜f (ph , qh ) is flatter than the original pf (ph , qh ); i.e.
@ p̃f
@ph
<
@pf
;
@pf
see
E NE
the Appendix for a proof of this result. Since by definition p˜f (pN
h , qh ) =
E NE
pf (pN
h , qh ) and the quality constraint will not bind below Ep , the new price
reaction function becomes P̃f Ep P˜f , instead of Pf Pf .
Let us suppose that before the shift to NFC the equilibrium price is at Ep .
As the home firm increases its price ph , the price of the foreign firm pf
will be less responsive to this change if it cannot change its quality beyond
qfN E . This is due to the high transport cost, and is illustrated by the locus
Ep P˜f . This is because any increase in own price brings additional benefit to
di↵erentiation and, without di↵erentiation, a firm will not increase its price
any further. The home firm knows that the foreign firm’s price response to
any given domestic price ph must keep its quality fixed. It can now maximize
profit by setting its price where its isoprofit IP is tangent to P̃f Ep Pf . The
home firm is now the price leader who maximizes profit, taking into account
the reaction function of the other firm:
⇧h (ph , qh ; Q̄h ) = ph Dh (p̃f (ph , qh ), ph , Q̄f , qh )
19
Ch (qh ).
(13)
The following first order conditions along with equation (12) define the equilibrium after the rise in NFC production, depicted in Figure 2 as Ẽp .
@ p˜f
+ Dphh ) + Dh (p, q) = 0
@ph
@ p˜f
= ph (Dphf
+ Dqhh ) Ch0 (qh ) = 0,
@qh
⇧hph = ph (Dphf
(14)
⇧hqh
(15)
This new equilibrium allows the home firm to di↵erentiate further and sell
its product at a higher price. From equation (14) and (15), it can easily be
shown that the price and quality of the home firm will be higher as long
as
@ p̃f
@ph
> 0 and
@ p̃f
@qh
< 0. Since the isoprofit IP at Ẽp is now higher than
what it would be at Ep , the home firm enjoys more profit at a higher level of
quality.9 More strikingly, the foreign firm will also make more profit, since it
can sell at a higher price even at the same quality. This new level of prices is
perceived by both firms as a credible strategy, since it is in the foreign firm’s
best interest if it cannot move toward NFC; i.e., this price lies on its best
response function. In the absence of a binding quality constraint, neither firm
would trust the other to raise the price above the pre-NFC level at Ep , even
though both would like higher prices. However, high transportation costs to
ship NFC to the U.S. market precommits the quality sold by the foreign firm
so that higher prices for both firms can be sustained. Both firms capitalize
on an increasingly segmented market as their products are less substitutable.
They both produce at the more inelastic portion of demand and extract more
economic rents at the expense of consumers.10 As the gap widens between
20
the perceived qualities that consumers perceive for NFC and reconstituted
FCOJ, this e↵ect will be greater.
We describe a simulation model in the next section to verify that the U.S.
production shift toward NFC juice may have rewarded both home and foreign firms. The prices received by both firms are generally higher and the
quality of the foreign firm’s product exported to the U.S. remains essentially
unchanged. The simulation model also shows that the change in profits is
more dramatic as consumers increase their quality awareness.
4
Simulated Trade Patterns
We investigate the theoretical result with a simulation of two partial equilibria, using price and sale data obtain from AC Nielsen (2005) along with
elasticities estimated by Brown and Lee (2000). We assume a log-log demand
function where the slope coefficients indicate own and cross elasticity of price
and quality;
lnDf = ↵h
⌘f lnpf + lnph +
lnDh = ↵f + lnpf
⌘h lnph
f lnqf
⇢lnqh
(16)
⇢lnqf +
h lnqh .
(17)
The cost of quality is assumed to be quadratic and identical across firms,
Ci (qi ) = kqi2 , for i = f, h. Brown and Lee (2000) estimated own and cross
21
price elasticities of NFC and FCOJ using a Rotterdam model. Table 3 reports
their elasticity estimates along with standard errors. The FCOJ and NFC
cross price elasticities were restricted to be the same.
[Table 3 about here.]
We calibrate the intercept parameters ↵f and ↵h such that the solution of
the system of equations, defined by equation (12), (14) and (15), yields the
FCOJ and NFC price and sales data obtained from AC Nielsen (2005).11
With the calibrated parameters, the equilibrium before the shift to NFC,
defined by equation (4) to (7) is then calculated.
The price elasticities in Table 3, when evaluated at AC Nielsen price and
quantities, implies that the slope parameters ⌘f , ⌘h , and
are 0.27, 0.30,
and 0.17, respectively. The own and cross quality elasticities defined by the
slope parameters
f,
h,
and ⇢ are unknown. Thus, we perform sensitivity
analysis over a reasonable range of parameter values. Due to the concavity
assumption in the profit maximization problem, i must be smaller than
p
2 ⌘i , for i = f, h. Thus, the upper bound of f and h , for a given ⌘f = 0.27
and ⌘h = 0.30, are 1.045 and 1.088, respectively. Since we are not considering
the case where quality declines (see Appendix C), for each considered value of
h,
the parameter ⇢ can not exceed
2⌘f
(4⌘f ⌘h
cross price elasticities, we calculate
2⌘f
(4⌘f ⌘h
2)
h.
2)
For given values of own and
= 0.3046. The upper bound
of ⇢ is then 0.3046*1.048=0.32. The intercept parameters ↵f and ↵h are
calibrated, for each of the varying unknown parameters, so that the solution
22
of the system of equation equals the 2005 prices and quantities. The values
and ranges of market parameters are summarized in Table 4.
[Table 4 about here.]
Table 5 reports the maximum and the minimum percent changes in price,
quality and profit of both home and foreign firms, for the entire range of own
and cross quality elasticities. Based on these results we conclude that the
simulation results agree with the theoretical model; both home and foreign
firms may gain higher profit through strategic behavior, even at the minimum price changes. Given the 2005 price and quantity data, along with the
elasticity estimates from the literature, our model suggests that the home
firm increases its quality by at least 5.06% and enjoys at least 0.85% higher
profit, due to the strategic shift towards NFC. The price of the home product
increases by at least 10.13%. The foreign firm produces at the same level of
quality but at a higher price, which is at least 1.23% higher. The profit of
the foreign firm increases by at least 8.41%.12 This provides a rough estimate of the strategy-induced price impacts on NFC. NFC consistently prices
at a premium of $1.50/gallon or roughly 30% of NFC price (AC Nielsen,
2005). Our estimates show that at least 10.13% of the price di↵erential can
be attributed to the strategic behavior of both firms.
[Table 5 about here.]
We further examine the change in profits of both firms by plotting graphs
across the values of own and cross price elasticities,
23
i
and ⇢. Figure 4
illustrates the results. It is clear that as long as the market parameters are
within a reasonable range, both home and foreign firms gain higher profits
as production shifts toward a new di↵erentiated product; all percent change
in profits are well above zero for both the home and foreign firms. A higher
cross quality elasticity (⇢) indicates stronger substitution between the two
products. Both firms would be less strategic in determining their optimal
qualities if consumers were willing to always substitute the quality of one
product for another. The top panel in Figure 4 shows that the percent
change in profits declines when the cross quality elasticity rises. Both firm
would find it more difficult to extract economic rents from consumers when
they perceive FCOJ and NFC as highly substitutable. A large own quality
elasticity ( i ) indicates a high degree of consumer quality awareness. The
bottom panel in Figure 4 shows that the change in profit becomes more
dramatic as consumers are highly sensitive to changes in product quality of
both goods. However, the own quality elasticity of the foreign firm ( f ) does
not a↵ect the change in profit of the home firm as indicated by the flat plane
in the direction of
f.
This is because the quality of the foreign firm is fixed
after the shift in U.S. production to NFC and the home firm is una↵ected
by the change in perception toward quality of the foreign firm. On the other
hand, as
f
increases the foreign firm’s profit before the shift to NFC declines,
while it remains fixed after. As a result, the change in the foreign firm’s profit
rises.
[Figure 4 about here.]
24
5
Conclusion
This paper demonstrates how high transport costs for a higher quality version
of a good that is both imported and produced domestically may give the
domestic industry an opportunity to specialize in the higher quality product.
The foreign producer does not follow the decision of the home firm to increase
quality, and instead implicitly cooperates, specializing in the lower quality
product. This provides impetus for product di↵erentiation by the domestic
industry, brought on by relatively large transport costs for the high quality
good.
The case of orange juice produced in Florida and imported from Brazil
demonstrates this finding. Brazil is the low-cost producer, but is less competitive in the U.S. market for fresh NFC juice, given the considerably higher
transport costs associated with the product. As a result, Brazil ships primarily FCOJ to the United States. In contrast, Brazil exports largely the
higher quality NFC to the European market, where the domestic industry
is not significant as in the U.S. The significant demand for NFC from U.S.
consumers shows that this di↵erence between export markets is not purely
driven by consumer tastes for the same product in di↵erent forms.
Both theoretical results and simulations indicate that a significant portion of
the premium paid by U.S. consumers for NFC is due to strategic adjustments
made by producers to move to this di↵erentiated equilibrium. We postulate
25
that the shift to NFC production in the U.S. was partially motivated by the
2004-2005 hurricanes in Florida that reduced the size of the crop by almost
50%. Rather than losing from its inability to follow Florida, Brazil also gains
from the increased specialization, and U.S. consumers lose.
The theoretical model described here has wider applications as it draws out
implications for other agricultural products, for example. Even when the
form of the product does not change, the same phenomenon exists with
attempts to obtain enforceable geographic indicators, such as wine and cheese
appellations. Out model assumes infinite transport cost to stylize the result.
In future work this assumption can be relaxed to investigate to which degree
the transport cost would a↵ect di↵erentiated quality, by explicit modeling
transport costs for each variety of good.
26
Notes
1
Markets in which the HHI is between 1000 and 1800 points are considered
to be moderately concentrated, and those in which the HHI is in excess of
1800 points are considered to be concentrated. Transactions that increase the
HHI by more than 100 points in concentrated markets presumptively raise
antitrust concerns, under the Horizontal Merger Guidelines issued by the
U.S. Department of Justice and the Federal Trade Commission. See Merger
Guidelines 1.51. (http://www.justice.gov/atr/public/testimony/hhi.
htm)
2
Muraro and Spreen (2003) estimate that production costs of FCOJ are
45 cents per gallon Single Strength Equivalents (SSE) in Brazil versus 75
cents per gallon SSE in Florida. Transportation cost and the Florida equalization tax add an additional 10 cents per gallon SSE. With a tari↵ of 30
cents, producers from both countries face roughly the same e↵ective cost of
production at 85 cents per gallon SSE. Meanwhile, the transport cost of NFC
is estimated to be 77 cents per gallon SSE. Even though the U.S. import tari↵
on NFC is 17 cents, smaller than that of FCOJ, Florida still has a significant
cost advantage producing NFC juice, because the transport cost is higher by
more than seven-fold when compared to that of FCOJ.
3
The degree of imperfection is not crucial, but it seems reasonable to
assume that at least some consumers care about country of origin or a more
27
familiar label, even when the product is homogeneous. For decades, for
instance, Florida producers have funded generic advertising of Florida juice,
not all juice. This is in contrast to generic advertising of beef or milk that
typically does not emphasize the country of origin.
4
This cost specification is adopted to ensure an interior solution for prod-
uct quality. One can assume either a convex cost function with a demand
function that is linear in quality, or a demand function that is concave in
quality with linear cost. For instance, if di↵erentiation were accomplished
by advertising, and not by transforming the product, we might expect the
cost to be linear, while increased advertising e↵ort brings diminishing incremental demand shifts. However, we might expect demand to be linear in
quality, which is increasingly costly to improve. An interior solution means
that neither of the principles of maximal or minimal product di↵erentiation
holds (Motta, 1993).
5
Many of the theoretical models of quality choice in the international
trade literature fall into the category of simultaneous choice models (Herguera et al., 2000). See Falvey (1979); Santoni and Van Cott (1980); Das
and Donnenfeld (1987, 1989) among others. On the other hand, the sequential choice model has received much attention because it is more realistic in
certain industries, such as automobiles. See Gabszewicz and Thisse (1979),
Shaked and Sutton (1982), Motta (1993), and Aoki and Prusa (1997), among
others.
28
6
In this undi↵erentiated case, the conjectural variation of firm i for firm
j is such that
7
@pi
@pj
=
@pi
@qj
=
@qi
@pj
=
@qi
@qj
= 0.
By replacing qi by advertising expenditure, A, dividing the top and the
bottom, we obtain
Dpi i
i
DA
=
Di
0
CA
which is
✏p
✏A
=
1 A
0 pD i
CA
=
A
,
pD i
since CA = 1; the
ratio of two elasticites with respect to own price and advertising equals the
share of advertising expenditure in total sales. The marginal cost of advertising is one since we express it in terms of total expenditure on advertising.
The only innovation here is that the condition is now a function of price and
quality of the other firm.
8
It can also be shown that all of the following results, with more restrictive
assumptions, will hold even with nonlinear demand. The linear demand
simply ensures a global maximum of the profit function.
9
The quality chosen by the home firm does not always increase. Either
for a very small Dqhh or a very large Dqfh , the quality actually decreases.
Nonetheless, the same result holds as both firms still gain more profit. As
we rule out this possibility in the empirical section, we argue that this is an
uninteresting result (see the Appendix for proof and discussion).
10
We can easily show this analytically if we assume non-zero marginal cost
of production. With non-zero marginal cost, one can derive a market power
expression
P MC
P
and show that it is decreasing as the equilibrium moves
from Ep to Ẽp ; i.e., own price elasticity declines.
29
11
We use weekly AC Nielsen price and quantity data for the last week
of 2005. The price of FCOJ and NFC was $3.73/gallon and $5.32/gallon,
respectively. The FCOJ and NFC quantities sold were 7.40 million gallons/week and 6.86 million gallons/week, respectively.
12
The maximum percentage change should be interpreted with caution,
since it corresponds to the parameter values that are very close to making
the Hessian matrix of the profit maximization problem singular.
30
References
Alchian, A.A., and W.R. Allen (1967) University Economics (Wadsworth
Pub. Co.)
Aoki, R., and T.J. Prusa (1997) ‘Sequential versus simultaneous choice
with endogenous quality.’ International Journal of Industrial Organization
15(1), 103–121
Bauman, Y. (2004) ‘Shipping the good apples out: a new perspective.’ Economic Inquiry 42(3), 534–536
Borcherding, T.E., and E. Silberberg (1978) ‘Shipping the good apples out:
the alchian and allen theorem reconsidered.’ The Journal of Political Economy 86(1), 131–138
Braga, Carlos Alberto Primo, and Simao Davi Silber (1991) ‘Brazilian frozen
concentrated orange juice : the folly of unfair trade cases.’ Policy Research
Working Paper Series 687, The World Bank, May
Brown, Carolyn, Mark G. Brown, Jonq Ying Lee, and Vera Sodek (2009) ‘Citrus reference book.’ Florida Department of Citrus, Economic and Market
Research Department
Brown, M.G., and J.Y. Lee (2000) ‘A measurement of the quality of orangejuice consumption.’ Agribusiness 16(3), 321–332
Crozet, M., K. Head, and T. Mayer (1989) ‘Quality Sorting and Trade:
31
Firm-level Evidence for French Wine.’ The Review of Economic Studies
79(2), 609–644
Das, S.P., and S. Donnenfeld (1987) ‘Trade policy and its impact on quality
of imports.’ Journal of International Economics 23(2), 77–95
(1989) ‘Oligopolistic competition and international trade:: Quantity and
quality restrictions.’ Journal of International Economics 27(3-4), 299–318
Donovan, Jason, and Barry Krisso↵ (2004) ‘The u.s. orange juice industry
in the ftaa.’ Chapter 5 in “Agriculture and the Free Trade Area of the
Americas” Economic Research Service/USDA Washington DC. AER-827
Dorfman, R., and P.O. Steiner (1954) ‘Optimal advertising and optimal quality.’ The American Economic Review 44(5), 826–836
Eastman, H., and S. Stykolt (1960) ‘A model for the study of protected
oligopolies.’ Economic Journal 70, 336–47
Echazu, Luciana (2009) ‘Product di↵erentiation, firm heterogeneity and international trade: Exploring the alchian-allen e↵ect.’ Research in Economics 63(2), 95 – 101
Falvey, R.E. (1979) ‘The composition of trade within import-restricted product categories.’ The Journal of Political Economy 87(5), 1105–1114
Feenstra, R.C. (1988) ‘Quality change under trade restraints in japanese
autos.’ The Quarterly Journal of Economics 103(1), 131–146
32
Feenstra, Robert C., and John Romalis (2014) ‘International prices and endogenous quality.’ The Quarterly Journal of Economics 129(2), 477–527
Gabszewicz, J.J., and J.F. Thisse (1979) ‘Price competition, quality and
income disparities.’ Journal of Economic Theory 20(3), 340–359
Gross, J., and W.L. Holahan (2003) ‘Credible collusion in spatially separated
markets.’ International Economic Review pp. 299–312
Hallak, Juan Carlos (2006) ‘Product quality and the direction of trade.’ Journal of International Economics 68(1), 238–265
Harris,
R. (1985) ‘Why voluntary export restraints are’voluntary’.’
The Canadian Journal of Economics/Revue canadienne d’Economique
18(4), 799–809
Hart, Ezequiel (2004) ‘The united states orange juice tari↵ and the “brazilian
invasion” of florida.’ Master’s thesis, Tufts School of Law and Diplomacy
Herguera, I., P. Kujal, and E. Petrakis (2000) ‘Quantity restrictions and endogenous quality choice.’ International Journal of Industrial Organization
18(8), 1259–1277
Hummels, D., and A. Skiba (2004) ‘Shipping the good apples out? an empirical confirmation of the alchian-allen conjecture.’ Journal of Political
Economy 112(6), 1384–1402
Irwin, D.A., and P. Temin (2001) ‘The antebellum tari↵ on cotton textiles
33
revisited.’ The Journal of Economic History 61(03), 777–798
Krishna, K. (1989) ‘Trade restrictions as facilitating practices.’ Journal of
International Economics 26(3-4), 251–270
Lehmann-Grube, U. (1997) ‘Strategic choice of quality when quality is costly:
the persistence of the high-quality advantage.’ The RAND Journal of Economics pp. 372–384
Martin, S. (1996) ‘R & d joint ventures and tacit product market collusion.’
European Journal of Political Economy 11(4), 733–741
Motta, M. (1993) ‘Endogenous quality choice: price vs. quantity competition.’ The Journal of Industrial Economics 41(2), 113–131
Muraro, R.P., and T.H. Spreen (2003) ‘Comparative Marketing Costs for
FCOJ from Florida and Sao Paulo.’ Department of Food and Resource
Economics, Florida Cooperative Extension Service, Institute of Food and
Agricultural Sciences, University of Florida
Raith, M. (1996) ‘Product di↵erentiation, uncertainty and the stability of
collusion, london school of economics.’ Technical Report, STICERD Discussion Paper Series EI/16, October
Ronnen, U. (1991) ‘Minimum quality standards, fixed costs, and competition.’ The RAND Journal of economics pp. 490–504
Ross, T.W. (1992) ‘Cartel stability and product di↵erentiation.’ International
34
Journal of Industrial Organization 10(1), 1–13
Santoni, G.J., and T.N. Van Cott (1980) ‘Import quotas: The quality adjustment problem.’ Southern Economic Journal 46(4), 1206–1211
Scherer, F.M., and D. Ross (1990) Industrial market structure and economic
performance (Houghton Mi✏in Boston, MA)
Shaked, A., and J. Sutton (1982) ‘Relaxing price competition through product di↵erentiation.’ The Review of Economic Studies 49(1), 3–13
35
A
Signs of Derivatives of Reaction Functions
The following proofs will show
@pi
@pj
@pi
@qi
> 0, @q
< 0, @q
< 0 and
j
j
@qi
@pj
> 0. Implic-
itly di↵erentiate (2) and use Cramer’s Rule to rearrange terms.
@pi (.)
=
@pj
⇧ipi pj ⇧ipi qi
Dpi j
⇧iqi pj ⇧iqi qi
0
|H⇧i |
=
Dqi i
Ci00
|H⇧i |
since Ci00 > 0 and Dpi j > 0 and |H⇧i | =
=
2Ci00 Dpi i
Ci00 Dpi j
2Ci00 Dpi i
Dqi i 2
> 0,
2
Dqi i > 0; the Hessian is
negative definite.
@pi (.)
=
@qj
⇧ipi qj ⇧ipi qi
Dqi j
⇧iqi qj ⇧iqi qi
0
|H⇧i |
=
Dqi i
Ci00
|H⇧i |
=
Ci00 Dqi j
2Ci00 Dpi i
Dqi i 2
< 0,
since Dqi j < 0.
@qi (.)
=
@qj
⇧ipi pi ⇧ipi qj
2Dpi i Dqi j
⇧iqi pi ⇧iqi qj
Dqi i
|H⇧i |
=
0
|H⇧i |
since Dqi i > 0.
36
=
Dqi i Dqi j
2Ci00 Dpi i
Dqi i 2
< 0,
@qi (.)
=
@pj
⇧ipi pi ⇧ipi pj
2Dpi i Dpi j
⇧iqi pi ⇧iqi pj
Dqi i
|H⇧i |
=
0
=
|H⇧i |
Dqi i Dpi j
2Ci00 Dpi i
Dqi i 2
> 0,
QED.
B
Restricted Reaction Function is Flatter
Totally di↵erentiate (12) with respect to pf and ph gives the left hand side
of the following inequality.
@ p˜f
=
@ph
Dpfh
2Dpff
<
Cf00 Dpfh
2Cf00 Dpff
2
Dqff
=
@pf
@ph
Rearranging terms reveals an obvious comparison.
@ p˜f
=
@ph
Dpfh
2Dpff
Dpfh
<
(See Appendix A for the derivation of
2Dpff
2
Dqff
00
Cf
@pf
)
@ph
QED.
37
=
@pf
@ph
C
Comparative Static on Quality of Home
Firm
The first order conditions of the profit function (13) are
@ p˜f
+ Dphh ) + Dh (p, q) = 0
@ph
@
p˜f
= ph (Dphf
+ Dqhh ) Ch0 (qh ) = 0,
@qh
⇧hph = ph (Dphf
⇧hqh
where
@ p˜f
@ph
=
Dpfh
> 0 and
2Dpff
@ p˜f
@qh
Dqfh
=
2Dpff
< 0, since Dpff < 0, Dpfh > 0, and
Dqfh < 0. Thus, price of home firm unambiguously increases when compared
@ p˜
to the solution obtain from equation (4) to (7), since (Dphf @phf + Dphh ) is a
smaller negative compared to Dphh . However, the quality will increase if and
only if
qh
qhold
or
(4⌘f ⌘h
2⌘f
2)
⇢<
h
Dqfh
= h
Dqh
2C 00 Dpff Dphh
Dqfh Dqhf
C 00 Dpff Dqhf
!
<1
if the demand equation is defined as (16) and (17) and if
the cost function is quadratic. This provide an upper bound to ⇢ and a lower
2⌘
(4⌘ ⌘
2)
bound to h . That is, 0 < ⇢ < (4⌘f ⌘hf 2 ) h . Since |H⇧h | > 0, f2⌘hf
⇢<
p
h < 2 ⌘h . Note that ⇢ can be interpreted as the cross conjectural variation
of foreign firm’s price for the home firm’s quality.
As a price leader, there are two opposing e↵ects when the home firm decides
38
to increase its quality, the direct e↵ect and the conjectural-indirect e↵ect.
First, the home firm gain profit from increase in demand for quality. The
parameter
h
determines the degree of this direct e↵ect through change in
demand. Second, since the home firm is a Stackberg leader, it knows that as
it increases its quality, the price of foreign firm will fall. The home firm loses
profit from smaller demand through the decrease in price of foreign firm.
This conjectural-indirect e↵ect is captured by ⇢. The first e↵ect dominate
the second e↵ect as long as
h
is not so small and ⇢ is not so large and the
quality will unambiguously increase.
We argue that the indirect e↵ect should be small. Thus, we rule out the
possibility when the home firm can increase profit by reducing its quality
and keeping ”good faith” that foreign firm will reduce its price just enough
such that the home firm’s profit does not fall.
39
List of Figures
1
2
3
4
U.S. Consumption and Production Shares of NFC and FCOJ
Orange Juice . . . . . . . . . . . . . . . . . . . . . . . . . .
Orange Juice Imperfect Competition in Price and Quality . .
Marginal Cost and Revenue of Quality . . . . . . . . . . . .
Percent Change in Profit of Foreign and Home Firm as the
Own and Cross Quality Slopes Vary . . . . . . . . . . . . . .
40
. 41
. 42
. 43
. 44
Figure 1: U.S. Consumption and Production Shares of NFC and FCOJ Orange Juice
70%
FCOJ Production
NFC Production
60%
50%
40%
30%
20%
10%
0%
1996-97 1997-98 1998-99 1999-00 2000-01 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11
Source: AC Nielsen (2014) and Economics and Market Research Department, Florida
Department of Citrus (FDOC-EMRD, 2014)
41
Figure 2: Orange Juice Imperfect Competition in Price and Quality
pf
Ph
IP
qf
IP
Pf
Qh
P̃f
Ẽp
Ep
Qf
P̃f
Q̄h
Eq
Pf
Ph
ph
42
Qh
Qf
qh
Figure 3: Marginal Cost and Revenue of Quality
M R, M C
f
C1 (q1 )
p1 Dq11
e
o
q1N E = Q̄1
43
q1
Figure 4: Percent Change in Profit of Foreign and Home Firm as the Own
and Cross Quality Slopes Vary
Foreign Firm
Foreign Firm
Home Firm
Home Firm
13
h
3
12
%
f
%Change
% in Profit
14.5
14
2
1
11
Cross Quality Elasticity rho
Cross Quality Elasticity rho
0.11
⇢
0.22
0.33
80
8
60
h
0
0
6
%
f
%Change
% in Profit
10
4
40
20
2
0
1
0
1
1
0.5
f f
Delta
0
0.11
⇢
f f
Delta
Delta hh
0.22
0.33
1
0.5
0.5
0
0
0.5
0
0
Delta hh
Note: All the percent changes are calculated by comparing two equilibria: before and after
the NFC shift. A simulation is performed by solving the system of equations with 100
repetitions for all possible parameter values 1 , 2 , and ⇢. In the top panel, the solutions
are computed based on the median of 1 at 0.53 and 2 at 0.54 while varying ⇢ 2 [0.00, 0.32].
In the second column, we fix ⇢ at its median value of 0.17 and vary 1 2 [0.00, 1.045], and
2 2 [0.00, 1.088] simultaneously.
44
List of Tables
1
2
3
4
5
Concentration in the Orange Juice Processing Industry in 2010
Shares of Brazil Exports by Orange Juice Category: Fresh
(NFC) versus Concentrated (FCOJ) . . . . . . . . . . . . . . .
U.S. Orange Juice Market Elasticities . . . . . . . . . . . . . .
Market Parameter Values and Possible Ranges . . . . . . . . .
Comparison of Equilibria Before and After Shift in Consumption to NFC . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
46
47
48
49
50
Table 1: Concentration in the Orange Juice Processing Industry in 2010
Processing Firms Based in United States
Capacity1
Firm
Market Share2
40
25-30
20-30
20
20
20
20
3
175
Tropicana
Cutrale
Peace River Citrus
Citrosuco
Louis Dreyfus
Southern Gardens/Florida’s Natural
Citrus World
Other
Total
25%
16%
14%
11%
11%
11%
11%
2%
100%
1,499
HHI
Processing Firms Based in Brazil
Firm
Cutrale
Citrosuco/Citrovita3
Louis Dreyfus
Total
Capacity1
Market Share2
135-150
105-120
37.5-52.5
300
47.5%
37.5%
15%
100%
3,888
HHI
1 Units
in millions of 90 lb. boxes
2 Shares
are calculated from midpoint of ranges
3 Citrosuco
and Citrovita agreed to merge on a 50-50 basis on May 14, 2010
Source: Compiled by authors through industry interviews
46
Table 2: Shares of Brazil Exports by Orange Juice Category: Fresh (NFC)
versus Concentrated (FCOJ)
Percent of Total Export
Destination
2004
2005
2006
2007
2008
2009
2010
2011
2012
US
NFC
FCOJ
0.0
99.9
0.0
99.9
0.0
99.7
10.7
89.3
12.3
87.7
26.0
74.0
24.7
75.3
36.3
63.7
21.4
78.7
EU
NFC
FCOJ
26.9
73.1
31.5
68.5
32.7
67.3
33.7
66.3
41.8
58.2
68.9
33.1
71.9
28.1
71.2
28.7
62.3
37.7
Source: Global Trade Atlas (2014)
47
Table 3: U.S. Orange Juice Market Elasticities
Price
FCOJ
NFC
FCOJ
-1.041
(0.036)
NFC
0.083
0.030)
0.083
(0.030)
-1.483
(0.052)
Source: Brown and Lee (2000),
calculated at sample mean budget share values (Table 4).
48
Table 4: Market Parameter Values and Possible Ranges
Source
Foreign
Home
Table 3
⌘f
0.27
Sensitivity Analysis
f
0.17
⌘h
0.30
[0.00, 1.045]
Calibration
⇢
[0.00, 0.32]
h
[0.00, 1.088]
49
↵f
varies
↵h
varies
Table 5: Comparison of Equilibria Before and After Shift in Consumption to
NFC
⇢ 2 [0.00, 0.32]
min
%
%
%
%
max
f
h
2 [0.00, 1.045]
2 [0.00, 1.088]
min
max
pf
ph
qf
qh
4.28%
11.90%
0.00%
6.77%
5.24%
13.97%
0.00%
13.97%
1.23%
10.13%
0.00%
5.06%
22.62%
31.60%
0.00%
25.55%
% ⇧f
% ⇧h
11.63%
1.21%
14.29%
1.52%
8.41%
0.85%
59.83%
6.73%
All the percent changes are calculated by comparing two
equilibria: before and after NFC shift. A simulation is performed by solving the system of equations with 100 repetitions for all possible parameter values. In the first column,
solutions are computed based on the median of f at 0.53
and h at 0.54 while varying ⇢ 2 [0.00, 0.32]. In the second column, we fix ⇢ at its median value of 0.17 and vary
f 2 [0.00, 1.045], and h 2 [0.00, 1.088] simultaneously.
50