“Impacto del Comercio Agrícola
con el MERCOSUR en el Estado de
Seguridad Alimentaria de China”
“Impact of Agricultural Trade with
MERCOSUR on the State of Food
Security in China”
Guillermo Larre
*
Daniel Gallardo
**
*
Guillermo Andrés Larre is a researcher in the Centre of Argentinian Studies in China
(CEAC), from the University of International Business and Economics (UIBE), in the Peo-
ple’s Republic of China, Beijing. His ORCID code is 0009-0003-3983-6219 and e-mail gui-
llermolarre@outlook.com
**
Daniel Alberto Gallardo is a researcher in the Centre of Argentinian Studies in China
(CEAC), from the University of International Business and Economics (UIBE), in the Peo-
ple’s Republic of China, Beijing. His ORCID code is 0009-0009-4858-9938 and e-mail da-
niel.gallardo90@gmail.com
http://dx.doi.org/10.22529/sp.2025.64.06
STUDIA POLITICÆ Número 64 primavera–verano 2025 pág. 153–174
Recibido: 03/03/2025 | Aceptado: 30/05/2025
Publicada por la Facultad de Ciencia Política y Relaciones Internacionales
de la Universidad Católica de Córdoba, Córdoba, República Argentina.
Abstract
This paper aims to show that the exports of agricultural commodities from
MERCOSUR countries to China has a signicant effect in the state of food
security in China. MERCOSUR countries have become key suppliers of
several agricultural commodities for China, especially soybeans and beef.
This suggests that MERCOSUR countries may have an important role in
154 STUDIA POLITICÆ Nº 64 primavera–verano 2025
the provision of proteins to the Chinese population. We focus on two di-
mensions of food security in China: availability and access. We use lineal
regression to show that the trade with MERCOSUR countries accounted
for 8.2% of the total protein supply, or 10.42 grams per person per day, in
mainland China, mainly due to trade in soybeans with Brazil. Then we use
a cointegration model to prove that the international prices of soybeans
are integrated across its main importers, China and the EU, and its main
exporters, Brazil, the USA, and Argentina. This shows that there is stable
connection between the international prices of this commodity, which has
an effect on costs associated with food price stability policies in China. Fi-
nally, we use an IRF to nd that Brazilian export prices of soybeans have a
delayed and permanent effect on Chinese import prices of that commodity.
A 1% increase in the soybeans exports price in Brazil can be expected to
produce a 0.32% increase in the import price in China after one month and a
2.8% permanent rise after approximately 8 months. This ndings will help
inform policymakers and scholars in the eld of food security and interna-
tional trade in China and MERCOSUR.
Keywords: agricultural trade, food security, soybeans, horizontal price
transmission, international passthrough
Resumen
Este artículo tiene como objetivo demostrar que las exportaciones de pro-
ductos agrícolas desde los países del MERCOSUR hacia China tienen
un efecto signicativo en el estado de la seguridad alimentaria en China.
Los países del MERCOSUR se han convertido en proveedores clave de
varios productos agrícolas para China, especialmente soja y carne vacu-
na. Esto sugiere que los países del MERCOSUR podrían desempeñar un
papel importante en el suministro de proteínas para la población china.
Nos enfocamos en dos dimensiones de la seguridad alimentaria en China:
disponibilidad y acceso. Utilizamos un modelo basado en regresión lineal
para demostrar que el comercio con los países del MERCOSUR explica el
8,29% del suministro total de proteínas, o 10,42 gramos por persona al día,
en China continental, explicado principalmente por el comercio de soja con
Brasil. Luego, utilizamos un modelo de cointegración para demostrar que
los precios internacionales de la soja están integrados entre sus principales
importadores, China y la UE, y sus principales exportadores, Brasil, Esta-
dos Unidos y Argentina. Esto muestra que existe una conexión estable entre
los precios internacionales de este producto, lo cual tiene un efecto en los
costos asociados con las políticas de estabilidad de precios de los alimentos
en China. Finalmente, utilizamos IRF para encontrar que los precios de ex-
portación de soja de Brasil tienen un efecto retardado y permanente en los
precios de importación de ese producto en China. Un aumento del 1% en
el precio de exportación de soja en Brasil puede esperarse que produzca un
aumento del 0,32% en el precio de importación en China después de un mes
y un aumento permanente del 2,8% después de aproximadamente 8 meses.
GUILLERMO LARRE Y DANIEL GALLARDO 155
Estos resultados ayudan a informar a las políticas y los estudios sobre se-
guridad alimentaria y comercio internacional en China y el MERCOSUR.
Palabras clave: comercio agrícola, seguridad alimentaria, soja, transmi-
sión horizontal de precios, traspaso internacional de precios
1. Introduction
T
his paper seeks to explain the impact that trade with MERCOSUR
countries has on the state of food security in the People’s Republic of
China, henceforth China. We will use descriptive statistics and analy-
sis based on linear regression and cointegration in order to quantify the pre-
cise effect that trading agricultural commodities with MERCOSUR countries
has had on food security in China. Our hypothesis is that agricultural trade
between MERCOSUR and China has reached such volume that it has gained
a signicant role in the state of food security in China, that we seek to iden-
tify and measure. We expect to help inform policymakers and scholars in the
elds of food security and international trade in both MERCOSUR countries
and China.
China is one of the world’s largest grain producers and is nearly self-suf
-
cient in grain production. According to the OECD-FAO Agricultural Out-
look report (2018), corn is the fastest-growing and most important cereal
produced in China, followed by rice and wheat. However, the production
of rice and wheat has shown slow growth in recent decades and is expected
to remain relatively stable in the coming decade due to a variety of factors,
while demand for animal feed is expected to continue growing. Thus, the
proportion of cereals used in animal feed will decrease as more soybean meal
is used, which relies on imports of raw soybeans. Between 2001 and 2018,
soybeans accounted for 75.4% of China’s total annual agricultural commodi-
ty imports, while rice and wheat together made up less than 6%.
China faces signicant challenges to its food security due to limited arable
land and freshwater resources. Urbanization and industrialization have con
-
tributed to a steady decline in arable land, with China’s per capita arable
land in 2020 measured at just 0.09 hectares per person–0.17 hectares below
the global average of 0.26 hectares. Similarly, China’s per capita freshwa-
ter availability in 2020 was only 2,239.8 m³, about one-fourth of the world
average. Despite these challenges, the country has managed to keep a high
156 STUDIA POLITICÆ Nº 64 primavera–verano 2025
level of food security and grain-sufciency. Monitoring data from the Natio-
nal Health Commission in China show that the average daily energy intake
per person in China has reached 2,172 kilocalories, comprising 65 grams of
protein, 80 grams of fat, and 301 grams of carbohydrates. The 2021 China
Agricultural Outlook Report (2022–2031) indicates that grain consumption
was 31.617 million tons, while actual grain production reached 66.234 mi
-
llion tons, with a ration self-sufciency rate of 197%. However, structural
imbalances persist. While staples like rice, wheat, and corn are often oversu-
pplied, high-quality wheat, specialty grains, and premium rice are undersu-
pplied and require imports.
Also, in the last decades, urbanization has shifted dietary preferences from
staple grains to protein-rich foods like meat, eggs, and dairy, creating a sig
-
nicant gap in the supply of protein feed grains. These imbalances between
production, supply, and consumer demand present risks to future food secu-
rity.
Over the past several years, maize and soybeans have played a crucial role in
China’s animal feed industry. Maize has been the dominant energy source for
feed due to its high carbohydrate content, while soybeans have provided the
essential protein necessary for poultry, pork, and dairy production. However,
the balance between these two crops in feed formulation has been shifting
due to changes in domestic supply, global trade relations, and government
policies.
Maize has remained the backbone of China’s feed sector. Historically, mai
-
ze demand outpaced production, leading to substantial imports, particularly
from the United States and Ukraine. However, since 2021, China has imple-
mented policies to increase domestic maize cultivation and reduce depen-
dency on foreign supply, leading to uctuating import volumes. Soybeans
have been heavily reliant on imports, such that the country has consistently
accounted for over 60% of global soybean imports. The following graph 1
shows the trends in agricultural imports in China:
GUILLERMO LARRE Y DANIEL GALLARDO 157
Graph 1: Agricultural imports in China, in tonnes, from 1961 to 2023
Source: FAOSTAT
We nd that soybeans, and to a much lesser extent maize, have come to do-
minate Chinese imports of agricultural commodities. Because of competition
over limited land, increasing demand for animal feed, and policies connected
to food security, China has come to depend on foreign markets for their su-
pply of soybeans. Furthermore, since 2018, trade tensions with the USA and
concerns about food security have driven China to diversify its suppliers,
relying more on Brazil while reducing purchases from the USA.
As mentioned, driven by rising consumption of meat, soybeans and mai
-
ze are experiencing continuing growth. Per USDA, in 2023, China produ-
ced approximately 277.2 million metric tons (MMT) of maize. To meet its
growing domestic demand, particularly for animal feed, China imported
about 27.14 MMT of maize in 2023, an increase of over 30% from 2022.
Notably, in 2023, China accounted for 18.2% of global maize imports, with
signicant purchases from Brazil and the United States.
Regarding soybeans, in 2023, China imported approximately 99.41 MMT,
marking a 9% increase from 2022. Domestic production also increased mo
-
derately, up to 19.7 MMT. That year, Brazil was China’s leading soybean su-
pplier, accounting for an estimated 77.3 MMT, which represents 74% of Chi-
na’s total soybean imports. The United States follows, supplying 21 MMT
and holding a 21% market share. The following graph 2 show the national
sources for the main agricultural commodities imported in China:
158 STUDIA POLITICÆ Nº 64 primavera–verano 2025
Graph 2: Agricultural imports of soybeans, maize, and beef in China by
origin country, in 2023.
Source: FAOSTAT
We can see that Brazil has become the biggest source of soybeans, maize,
and beef in China. Other signicant agricultural suppliers are the USA and
Argentina, while others still are signicant in individual items, like Ukraine
in maize, and Uruguay, Australia, and New Zealand in beef.
To ensure food security, Chinese policies aim to maintain absolute grain
self-sufciency while expanding trade channels with diverse import sources.
This strategy is particularly critical for soybeans and coarse grains, where
domestic production lacks comparative advantages.
We nd that today China has ensured its self-sufciency in cereals such as
wheat and rice. On the other hand, with respect to corn, it is beginning to
have certain difculties, which is why it is seeking to ensure supplies from
foreign markets, as is the case of the Argentine market, which in 2024 China
enabled imports from. A different case is in soybeans, a grain that is essential
for the supply of protein, not only directly for human consumption (through
industrial processes) but also increasingly essential for use as animal feed
that is then consumed by Chinese consumers. For this particular case, China
is heavily dependent on imports, especially from the United States, Brazil
and Argentina. Initially, the United States was the main supplier of soybeans,
however this changed radically, being surpassed by Brazil from the year
2013, a difference that was amplied after the so-called trade war between
the USA and China.
The MERCOSUR countries, with Brazil at its head, strongly increased ex
-
ports of other agricultural commodities to China. Currently, China is the
main destination for exports of soybeans, beef, pork and other smaller agri-
cultural commodities from MERCOSUR countries. The following table 1
shows recent volumes in agricultural trade between MERCOSUR and China:
GUILLERMO LARRE Y DANIEL GALLARDO 159
Table 1. MERCOSUR countries’ agricultural exports to China
Country Soybeans
(MMT)
Maize
(MMT)
Beef (Metric
Tons)
Other Key Commodities
Brazil 60-70 4-5 1,000,000+ Poultry, sugar, cotton
Argentina 6-8 1-2 500,000+ Soybean meal, soybean oil
Paraguay 5-6 0.5-1 100,000+ Wheat
Uruguay 2-3 <0.5 300,000+ Dairy, wool
Note: USDA estimates for the year 2021.
The trade volume of Brazilian exports to China is signicantly larger than the
other MERCOSUR countries. Their soybean exports to China have grown
from approximately 5-10 MMT annually in the early 2000s to over 60 MMT
in recent years, accounting for more than 70% of Brazil’s total soybean ex
-
ports; whereas their maize exports to China have surged in recent years, rea-
ching over 4 MMT annually by 2022, as China diversied its import sources
amid trade tensions with the United States. Brazil also exports signicant
quantities of beef, poultry, sugar, and cotton to China, with beef exports ex
-
ceeding 1 million metric tons annually in recent years.
On the other hand, Argentina is also a major exporter of soybean products
and maize, with a stronger focus on value-added products. Argentina is the
world’s largest exporter of soybean meal and oil, with China importing
approximately 10-15 MMT of soybean meal annually. Raw soybean exports
to China average 6-8 MMT per year; in addition, their maize exports to Chi
-
na have grown steadily, reaching 1-2 MMT annually in recent years; and it
also exports beef, pork, and dairy products to China, though in smaller volu-
mes compared to Brazil.
Next, Uruguayan soybean exports to China have grown signicantly, rea
-
ching 2-3 MMT annually in recent years. China is Uruguay’s largest beef
market, with exports growing from around 50,000 metric tons annually in
the early 2000s to over 300,000 metric tons in recent years. The country also
exports dairy products and wool to China.
Finally, Paraguayan exports to China are routed through other countries,
mainly Argentina, so there are no ofcial precise numbers on them, however
we may nd approximate estimates in USDA and other sources. They esti
-
mate that Paraguay’s soybean exports to China have grown from less than
160 STUDIA POLITICÆ Nº 64 primavera–verano 2025
1 MMT annually in the early 2000s to over 5 MMT in recent years; while
their maize exports to China have increased to 0.5-1 MMT annually; and that
Paraguay also exports beef and wheat to China, in modest volumes.
These increasing volumes of agricultural exports, mainly in soybeans, maize,
and beef, from MERCOSUR countries to China lend support to our hypothe
-
sis that these trade links have a role in determining the state of food security
in China, which we will elucidate with adequate empirical research.
2. Literature Review
Food security has had several contrasting denitions over time, but the most
widely accepted denition today is the one produced in the 1996 World Food
Summit that dened food security as a state “when all people at all times
have physical and economic access to sufcient, safe and nutritious food to
meet their dietary needs and food preferences for an active and healthy life.”
Thus, food insecurity exists when people do not have sufcient physical or
economic access to food. In contrast, food self-sufciency is dened as the
ability to meet consumption needs, particularly for staple foods, from a coun
-
try’s own domestic production rather than having to rely on importing or
buying from non-domestic sources (minimizing dependence on international
trade). Thus, food security exists irrespective of the domestic or international
source of the available food, whereas self-sufciency is centered on a coun-
try’s ability to provide food security for itself. The denition of food security
emphasizes four essential dimensions: availability, access, utilization, and
stability, which together form the foundation of food security. Each of the-
se dimensions highlights a critical aspect of what it means for a population
to be food secure. In this paper, we will focus on the effect that trade from
MERCOSUR has on the availability of food and its access, represented by
its price.
Our analysis of the effects of food trade on local prices is grounded in the
theoretical principles of the Law of One Price (LOP), rst introduced by
Marshall (1890), which is a cornerstone of economic theory. The LOP posits
that the price of a single product will be identical in two markets that enga
-
ge in trade, assuming no barriers such as tariffs or trade restrictions. This
price alignment occurs because market participants capitalize on any price
discrepancies in different markets to generate prot through trade. While lo-
cal supply and demand conditions may create temporary price differences
between regions, these gaps are eventually closed by market participants.
GUILLERMO LARRE Y DANIEL GALLARDO 161
Arbitrageurs purchase goods in lower-priced markets, reducing local supply,
and sell them in higher-priced markets, increasing supply there. This adjust-
ment process places upward and downward pressure on prices in the respec-
tive markets, ultimately driving them toward equality in the long run. For a
comprehensive discussion of the LOP, refer to Fackler and Goodwin (2001).
When prices are internationally transmitted by the effect of the LOP, this
is called price passthrough. Kenen and Pack (1980) provide an extensive
analysis of the passthrough of global prices to domestic markets, focusing
on imports. They attribute incomplete passthrough rates to demand elas
-
ticities, suggesting that unitary elasticity results in delayed, but full, price
transmission in the long run. In contrast, Kirchgassner and Kubler (1991),
also examining imports, argue that incomplete passthrough is inuenced by
market structure. This aligns with other studies that also link pricing power
to incomplete price transmission (Guvheya et al., 1998; Taylor, 2000; Miller
and Hayenga, 2001).
Liu et al. (2018) study price passthrough in China and nd that China’s Eas
-
tern regions, characterized by greater integration with international trade, are
notably inuenced by global market dynamics. The study reveals that price
uctuations in these areas are strongly shaped by trends in international mar-
kets. This heightened exposure to global trade contributes to increased price
volatility in the Eastern provinces, resulting in less stable price levels and
weaker convergence with those of other regions within China.
Focusing on soybeans, Coughlin and Sutton-Vermeulen (2020) use simple
correlations to analyze the price passthrough of this commodity in China
and other countries. Among other results, for the purpose of our study, they
nd that there is price transmission between Chinese and Brazilian soybean
prices, which strengthens during Brazil’s export season. The long-term con
-
nection between the Dalian Spot Cash price in China and a combined FOB
U.S. Gulf/FOB Brazil price indicates that, under normal market conditions
free of trade disruptions or supply/demand shocks, the Chinese benchmark
closely tracks prices from the dominant export origins, ie. American and Bra-
zilian prices during each of their harvest seasons. Over the past decade, the
correlation between Chinese soybean prices and seasonally-adjusted export
prices has averaged around 82%.
Other authors also provide evidence for a strong passthrough effect in soy
-
beans prices in Brazil, the USA, and China. Machado and Margarido (2001)
and Maoletti (2001) provide further evidence of rapid or instantaneous pri-
162 STUDIA POLITICÆ Nº 64 primavera–verano 2025
ce transmission in the soybean market, indicating high market efciency.
Finally, focusing on the price transmission dynamics in the global soybean
market during the US-China trade war, Barboza Martignone, Behrendt and
Paparasthe (2022) examined price relationships across key players in the
soybean trade, including China, the USA, the European Union, Brazil, and
Argentina. The authors apply various econometric methods, including co
-
integration analysis, to evaluate how the trade war inuenced market inte-
gration and efciency. Their ndings reveal that the global soybean market
remains highly efcient and cointegrated, with prices across regions moving
together over time. Despite disruptions caused by the trade war, such as tari-
ffs and subsidies, the market demonstrated a high degree of integration and
price transmission, reafrming the validity of the LOP in the long term. The
study emphasizes the international soybean market’s ability to adjust and
realign prices and maintain market efciency, even during periods of geopo
-
litical instability.
We nd that the literature lends support to our hypothesis that MERCOSUR
countries may hold inuence over the price of food commodities in China.
We can expect Brazil to have a signicant, measurable effect in the prices of
soybeans in China.
3. Empirical Models
We will approach the question of the effect of agricultural trade with MER
-
COSUR on Chinese food security from two different angles. On the one
hand, we will attempt to quantify precisely how much of the protein intake of
the Chinese population can be attributed to the agricultural trade with MER-
COSUR. For this, we will use a linear regression model, which will consider
the effect of the total supply of maize and soybeans on the protein supply
in order to produce conversion rates of these commodities into the nutrition
of the Chinese population. Then we will apply those conversion rates to the
maize and soybeans quantities traded from MERCOSUR countries to China,
in order to produce the amount of protein that MERCOSUR countries supply
to the people of China, by way of trade of soybeans and maize.
Next, we will attempt to identify the inuence that the agricultural trade with
MERCOSUR countries has had on food prices in China. For this, we will use
a cointegration model that puts together the international prices of the major
exporters of soybeans, that is Brazil, the USA, and Argentina, with those
GUILLERMO LARRE Y DANIEL GALLARDO 163
of the major importers, China and the EU. The purpose of this model is to
identify price transmission between Brazil and Argentina on the one side and
China on the other. One limitation of this model is that it does not connect
the imports price in China to the local domestic price accessible to the Chi
-
nese population, which is a better measure to study food security. However,
internal prices are heavily regulated in China, so it is likely such a model
would yield little useful information. Instead, this model will help elucidate
the connection between the international prices of soybeans in MERCOSUR
and China, which has a role in the costs associated to Chinese policies on
price regulation. This kind of cointegration model could be applied to maize
and beef as well, not just soybeans, however these commodities are exported
in relatively small quantities by MERCOSUR countries to China, and this
precludes the possibility of a signicant inuence by way of trade.
The linear regression model will be used to quantify the effect that the avai
-
lability of soybeans and maize in China has on the protein supply in that
country. It is grounded on the fact that these two crops constitute the vast
majority of animal feed used in China, which in turn is the main source of
protein for the Chinese population. Once we quantify the effect of maize and
soybeans in terms of protein supply, or the conversion rates for these crops
into protein supply, we may estimate the effect of MERCOSUR exports of
maize and soybeans to China by way of increased supply.
The rst step is thus to outt a model that may clarify the relation between
total availability of these crops and protein supply in China. The dependent
variable is protein supply as reported by FAO, available in series of 3-year
moving averages centered in the years 2000 to 2022. For the independent
variables, total availability of maize and soybeans will be calculated by ad
-
ding the local Chinese production together with the imports of these crops in
China and subtracting that country’s exports.
Once the time series for total available maize and soybeans in China are
produced, they can be tested for stationarity. We expect these series not to be
stationary but to contain unit roots. If they do, the next step is to take the rst
difference of these series in order to make them suitable for incorporation in
models of lineal regression analysis.
The last variable to include will be a control variable for the economic grow
-
th of China, which can also have an effect in protein supply by way of other
imports and production of minor sources of protein. The control variable
chosen for this purpose is GDP per capita of China.
164 STUDIA POLITICÆ Nº 64 primavera–verano 2025
It is important to note that the relatively low amount of observations (23)
does not allow for the incorporation of more commodities or control varia
-
bles that might make for a more rened model. Still, these variables sufce
for our purpose. The model produced then takes the form:
The model will then tell us the effect that total available maize and soybeans
in China have on the supply of proteins, measured in grams per person per
day, in that country. β
1
and β
2
are thus the conversion rates between maize
and soybeans to protein uptake in China. Once they have been calculated
we may multiply β
1
by the amount of soybeans exported from MERCOSUR
countries to China in order to calculate the exact amount of protein supply
in China that can be explained by the trade of soybeans with MERCOSUR.
We may do similarly with β
2
to calculate the same for maize. We may then
add the results together to nd the total amount of protein supply in China
that is generated from trade in agricultural commodities with MERCOSUR
countries. This is valuable input to inform our understanding of the effect of
agricultural trade with MERCOSUR in the food security in China.
The results are likely to underestimate the true value, because they do not in
-
clude the direct trade of beef and soybean meal from MERCOSUR countries
to China, which are other, more direct, sources of protein. We will assume
that this underestimation is not signicant, owing to their low ratio to local
production. We did not run this model on soybeans meal or beef because
their traded amounts are not as signicant, and because the effect of soybeans
and maize into protein supply occurs partly through these other variables,
such that their inclusion would yield misleading results for soybeans and
maize. Furthermore the data set is relatively short, so it is preferable not to
include too many variables.
The data set was obtained from FAOSTAT and it describes the production,
imports and exports of soybeans and maize, in tonnes, in China from 1961 to
2023, as well as additional data for protein supply and GDP per capita.
The other model included in our empirical research is a cointegration model.
The purpose of the cointegration model is to test the effect of the exports of
agricultural goods from MERCOSUR countries into China in terms of the
international price of soybeans in China. We will analyze the transmission of
price information in soybeans from the MERCOSUR countries into China,
which will reveal whether there is an inuence of Brazilian and/or Argenti
-
GUILLERMO LARRE Y DANIEL GALLARDO 165
nian on Chinese soybeans prices. Per the LOP, we should expect this effect
to be statistically signicant and measurable.
For this purpose, we will t a Vector Error Correction Model (VECM) model
including the FOB prices of the major exporters of soybeans, Brazil, Argen
-
tina, and the USA, and the CFR prices of the major importers, China and
the EU. Models of cointegration, like VECM, are extensions of VAR models
that attempt to identify a long-run equilibrium connecting the variables under
study, as well as short-term correlations between them. The long-run trend
is represented by one or more cointegrating vectors, that are produced as the
residuals of a lineal regression of the level form of the variables. Despite the
variables being non-stationary in level form, the residual does not carry a unit
root as long as the variables are cointegrated, or follow a similar long-term
trend. The vector of residuals is then included into a VAR together with the
variables in rst difference form. By taking the rst difference, the unit roots
are removed from them. Thus, the cointegrating vectors and the variables in
difference form are all made stationary, and they may be analyzed with stan
-
dard lineal regression techniques. In this VAR model, the cointegrating vector
informs about long-run connections while the differenced variables inform
about short-run correlations among the variables included in the model.
Thus, for our purposes, the VECM would take the form:
Since this is a VAR conguration, the variables in the formula are actually
vectors of variables and the Greek letters stand for vectors or matrices of
coefcients:
• PRICE is a vector of the ve time-series of prices of soybeans employed:
Chinese CFR prices, European CFR prices, Brazilian FOB prices, Ame
-
rican FOB prices, and Argentinian FOB prices.
• ECT is the error correction term, or the cointegrating vector that captures
the long-run equilibrium among the prices.
• Π is a 5×1 vector of coefcients of long-run adjustment, indicating how
each price responds to deviations from equilibrium.
• Г is the matrix of short-run coefcients, measuring correlations between
differenced price variables for a set number of lags.
• ε is the vector of error terms.
166 STUDIA POLITICÆ Nº 64 primavera–verano 2025
For the VECM model to be stable, or to represent a long-run relationship in
equilibrium, the coefcients in the Π vector must be signicant and have
a negative sign, which represents that variables in the long-run tend to fall
back into each other, or that they do not diverge permanently. This is becau
-
se the ECT represents a fraction of their difference, thus the variables must
behave in the opposite direction that the ECT is taking, in order for them to
sustain a long-term equilibrium.
For this model, we will use the log form of the price variables. This is becau
-
se it will allow us to read the resulting coefcients as elasticities. We must
consider this when interpreting the results. The short-run coefcients in the Γ
matrix will represent percentage changes, or the way price changes correlate
with each other within the specied lag length.
The VECM will allow us to prove that there is price transmission in the long
run, if all the prices included in the model have signicant, negative Π coe
-
fcients; and furthermore, it will reveal whether there is any short-term trans-
mission occurring between Brazil and Argentina on the one side and China
on the other. The USA and EU prices have to be included in the model or we
would incur a high chance of missing-variable bias. On the other hand, Uru-
guay and Paraguay were not included in the model because their relatively
small traded quantities preclude the possibility of a price transmission effect.
One following step can be undertaken to further clarify the relation between
the variables under study. An Impulse Response Function (IRF) can follow
a VECM. It is a tool used to analyze how a shock to one variable affects the
other variables over time. When a change, or shock, occurs in one variable,
the IRF traces the impact of that shock on all other variables in the system
across multiple time periods. This allows us to understand the transmission
of shocks and how long it takes for variables to stabilize.
The IRF is calculated by recursive substitution of a reduced-VAR form of the
VECM equation to produce an MA representation:
Where Ψ
j
is a matrix representing the effect of a shock in one price at time t
into all other prices at an arbitrarily selected number of lags, represented by
j. The width of this matrix is given by the length of the vectors of variables,
in this case 5, and its length is given by j. From this Ψ
j
matrix we may obtain
the response of one price variable to a shock originating in another price,
called the impulse.
GUILLERMO LARRE Y DANIEL GALLARDO 167
The interpretation of an IRF depends on both the magnitude and the direction
of the response. If a shock in one variable leads to a temporary change in ano
-
ther that gradually fades, this suggests a short-term effect without long-term
consequences. Conversely, if the response persists, it may indicate a more
fundamental relationship between the variables. In a VECM framework, this
distinction is particularly important because the model accounts for long-run
equilibrium relationships, meaning that some shocks may cause permanent
shifts, while others may be corrected over time. Finally, we may produce
an IRF graph which visually represents this dynamic adjustment process,
showing how the variables react period by period until stability is restored.
For the cointegration and IRF, we used international prices obtained from
USDA publications. All the models described will be resolved by use of
STATA 11.
4. Results
The ADF test for stationarity results show that all the variables to be used in
both models contain a unit root. We considered this when tting the models
for empirical analysis.
The results of the linear regression model are summarized in the following
table 2:
Table 2. Linear regression results
Coef. Std.
Error
t P>z 95% Condence Interval
ΔMaize 0.00000
00740
0.00000
00353
2.09 0.050
*
0.00000000
00650
0.000000
148
ΔSoy
-
beans
0.00000
0142
0.00000
00676
2.10 0.050
*
0.00000000
0143
0.000000
284
ΔGDP_
per_cap
0.00347
66
0.00113
75
3.06 0.007
***
0.0010868 0.0058664
P>F 0.0001
Adj. R2 0.6162
Note: The dependent variable is ΔProtein_Supply. Stars for level of signicance.
168 STUDIA POLITICÆ Nº 64 primavera–verano 2025
The results tell us that for every tonne of available maize in mainland China
during the period 2000 to 2022, protein supply increased by 0.0000000740
grams per person per day, whereas soybeans available in mainland China in
this period increased protein supply by 0.000000142 g/cap/d.
We may use these numbers multiplied by the quantities of imported soybeans
from MERCOSUR to obtain the amount of protein supply in mainland China
that can be attributed to the trade of maize and soybeans with these countries.
These results are summarized in the following table 3:
Table 3. Protein supply
Source Country Crop Imports Quantity
(tonnes)
Protein Supply
(g/cap/d)
Brazil Soybeans 64277600 9.127
Maize 71.98 0.000005
Brazil total 9.127
Argentina Soybeans 7455650 1.058
Maize 0.3 0.00000002
Argentina total 1.058
Uruguay Soybeans 1656570 0.235
Maize - -
Uruguay total 0.235
Soybeans total 73389820.08 10.421
Maize total 72.28 0.000005
MERCOSUR total 10.421
Note: Maize and soybeans imports quantities, protein supply levels, and percentage over
total protein supply. Year 2020.
We may additionally use those results to consequently calculate the percenta-
ge over total protein supply that these amounts represented for the year 2020.
The results show that:
GUILLERMO LARRE Y DANIEL GALLARDO
• The trade of soybeans and maize with MERCOSUR countries explains
8.29% of the total protein supply, or 10.42 grams per person per day, in
mainland China in the year 2020.
• These results are almost exclusively in the trade of soybeans, whereas
the trade of maize has a negligible effect.
• The main contributor is Brazil, with 9.12 g/cap/d or 7.26% of total pro
-
tein supply; followed by Argentina with 1.05 g/cap/d or 0.84% of the
total; and nally Uruguay with 0.23 g/cap/d or 0.18% of the total.
• Paraguay does not ofcially trade with China directly, its harvest is rou
-
ted mainly through Argentina, so some of the results attributed to Ar-
gentina in the model must actually come from Paraguay, but there is no
reliable way to make that distinction.
The following model, a VECM, connects the exports prices of Brazil, Ar
-
gentina, and the USA to the imports prices of China and the EU. It was ran
at 4 lags as per the results of an AIC test. The results for China are in the
following table 4:
Table 4. VECM results for China.
Coef. St. Err. z P>z 95% Condence
Interval
Coint. vector -0.025 0.015 -1.70 0.088* -0.055 0.003
ΔUSA - lag 1 0.204 0.038 5.30 0.000*** 0.128 0.28
ΔUSA - lag 2 0.058 0.041 1.39 0.165 -0.023 0.14
ΔUSA - lag 3 -0.085 0.040 -2.10 0.036** -0.166 -0.005
ΔBrazil - lag 1 0.048 0.044 1.09 0.274 -0.038 0.134
ΔBrazil - lag 2 0.066 0.042 1.57 0.116 -0.016 0.15
ΔBrazil - lag 3 0.136 0.043 3.15 0.002*** 0.051 0.221
ΔArgentina - l1 -0.006 0.049 -0.13 0.893 -0.104 0.091
ΔArgentina - l2 -0.034 0.049 -0.70 0.484 -0.13 0.061
ΔArgentina - l3 -0.073 0.051 -1.44 0.151 -0.174 0.026
170 STUDIA POLITICÆ Nº 64 primavera–verano 2025
ΔEU - lag 1 0.04 0.043 0.94 0.347 -0.044 0.125
ΔEU - lag 2 0.193 0.042 4.50 0.000*** 0.109 0.277
ΔEU - lag 3 0.073 0.044 1.66 0.097* -0.013 0.16
ΔChina - lag 1 0.166 0.059 2.78 0.005*** 0.049 0.283
ΔChina - lag 2 0.172 0.055 3.13 0.002*** 0.064 0.28
ΔChina - lag 3 -0.013 0.042 -0.32 0.747 -0.097 0.069
Note: Dependent variable is difference in logs of Chinese CIF price of soybeans. Stars for
level of signicance.
The results show that:
• Chinese prices are integrated with those of the rest of the world, speci
-
cally the USA, the EU, Brazil, and Argentina, which means that there
is a long-run balance that keeps these prices close to each other without
major permanent deviations. This is shown by the statistical signicance
and negative sign of the coefcient for the cointegrating vector for Chi-
nese prices as well as for the other prices in the model, not shown here.
• We nd that the inuence of Brazilian prices is not as strong as that of
American and European prices. Whereas these two regions are signi
-
cant in two out of the three lags considered, Brazil is only signicant
in one, and further their coefcients are higher than the Brazilian one.
This is surprising given the fact Brazil has become the largest exporter
of soybeans to China since 2013. It shows that American prices are still
used as reference prices by Brazilian and Chinese traders. On the other
hand, the connection with European prices makes sense, since the EU
and China compete directly for imports of soybeans, so the LOP is in full
effect between them.
• Furthermore, Argentinian export prices do not directly affect Chinese
prices. This can be explained by the relatively low quantities of Argen
-
tinian exports of this product compared to Brazil and the USA. It could
also be explained by the high degree of collinearity between Argentina
and Brazil, since their seasonal patterns are closely aligned, however
removing Brazil from the model does not affect these results.
Since we have established that Brazilian exports prices of soybeans do have
an effect over Chinese imports prices of this commodity, we may outt an
GUILLERMO LARRE Y DANIEL GALLARDO 171
Impulse Response Function (IRF) to gauge this inuence over time. Doing
so yields the results summarized in the following table 5:
Table 5. IRF results.
lags IRF Lower Upper
0 0.00161 -0.000349 0.00357
1 0.00329 0.000469 0.00612
2 0.00765 0.00366 0.0116
3 0.0147 0.00862 0.0207
4 0.0222 0.0137 0.0307
5 0.025 0.0143 0.0358
6 0.0265 0.0135 0.0394
7 0.0271 0.0123 0.0419
8 0.0279 0.0114 0.0443
9 0.0284 0.0104 0.0463
10 0.0286 0.0095 0.0478
Note: Impulse is changes to Brazilian FOB prices of soybeans, response is Chinese imports
prices.
The results show that:
• The immediate effect of changes in Brazilian prices is not signicant. We
see this because lag 0, representing immediate transmission, is not statis
-
tically signicant since its 95% condence interval includes the value 0.
• The effect is signicant starting with one lag, that is a one month diffe
-
rence, where a 1% increase in Brazilian prices can be expected to produ-
ce a 0.32% rise in Chinese imports prices.
• The effect grows exponentially through lags 2 and 3, then begins to taper
off and stabilizes after approximately 8 lags at a 2.8% rise in Chinese
prices. This tells us that a 1% increase in Brazilian exports prices of
soybeans can be expected to produce a permanent increase in Chinese
imports prices of this commodity of approximately 2.8% after 8 months.
172 STUDIA POLITICÆ Nº 64 primavera–verano 2025
This suggests a strong and persistent price transmission effect from Bra-
zil to China.
These results are visible in the following IRF plot:
Graph 3. IRF graph. Impulse is Brazilian prices and response is Chinese
prices
5. Conclusions
We have shown that Brazil has become a key supplier of every major agri
-
cultural commodity imported by China. This includes soybeans, maize, and
beef. Argentina is also an important supplier of soybeans and beef, and Uru-
guay is signicant in beef.
Furthermore, by use of descriptive statistics and empirical research based on
lineal regression and cointegration, we nd support for our hypothesis that
trade with MERCOSUR countries has a signicant role in the state of food
security in China. To be precise, our research shows that by the year 2020,
the trade with MERCOSUR countries explained 8.29% of the total protein
supply, or 10.42 grams per person per day, in mainland China, explained
mostly by the trade of soybeans. Trade with Brazil explains the majority of
this effect, or around 87% of the available proteins in China that can be attri
-
buted to agricultural trade with MERCOSUR countries.
Finally, Brazilian exports prices of soybeans have a direct impact on Chine
-
se imports prices of that commodity. This impact on prices is delayed, not
immediate, but it is permanent, starting at one month with a 0.32% rise in
Chinese imports prices after a 1% change in Brazil, and growing exponentia-
lly until it stabilizes after approximately 8 months at a 2.8% rise in Chinese
prices for a 1% change in Brazil.
Our research is based on ofcial data and robust econometric techniques. It
GUILLERMO LARRE Y DANIEL GALLARDO 173
is possible though that it underestimates the true amounts of protein available
in China owing to trade with MERCOSUR because it does not consider di
-
rect imports of beef and soybeans meal. Also, it does not consider the domes-
tic passthrough on food prices, but rather imports prices which only have an
effect on the costs associated with price stabilization policies, but not on food
security per se. These are limitations owing to the structure and size of the
data employed for the research. As agricultural trade continues and the data
set expands, we may address some of these limitations. Others will require an
expanded set of variables and a different approach to answer the question.
References
Barboza Martignone, G., Behrendt, K., & Paparas, D. (2022). Price transmission analysis of
the international soybean market in a trade war context. Economies, 10(8), 203.
Coughlin, A., & Sutton-Vermeulen, D. (2020). The relationship between global soybean
prices and local markets. CME Group. https://www.cmegroup.com/education/arti-
cles-and-reports/ relationship-between-global-soybean-prices.html
Fackler, P. L., & Goodwin, B. K. (2001). Spatial price analysis. In B. L. Gardner & G. C.
Rausser (Eds.), Handbook of agricultural economics (Vol. 1, pp. 971–1024). Elsevier.
Guvheya, G., Mabaya, E. T., & Christy, R. D. (1998). Horticultural marketing in Zimba-
bwe: Margins, price transmission, and spatial market integration (Staff Paper No. SP
98-02). Cornell University.
Kenen, P. B., & Pack, C. (2008). Exchange rates, domestic prices and the adjustment pro-
cess: A survey of recent evidence (Occasional Paper No. 1). Group of Thirty.
Kirchgässner, G., & Kübler, K. (1992). Symmetric or asymmetric price adjustments in the
oil market: an empirical analysis of the relations between international and domes-
tic prices in the Federal Republic of Germany, 1972-1989. Energy Economics, 14(3),
171–185.
Liu, T. Y., Su, C. W., Chang, H. L., & Xiong, D. P. (2018). Does the law of one price hold?
A cross-regional study of China. Economic Research-Ekonomska Istraživanja, 31(1),
592–606.
Machado, E. L., & Margarido, M. A. (2000). Seasonal price transmission in soybean inter-
national market: The case of Brazil and Argentina. Pesquisa & Debate, 12, 92–106.
Maoletti, R. L. (2001). Formação de preços na cadeia agroindustrial da soja na década de
90. Revista de Economia e Sociologia Rural, 39, 9–26.
Marshall, A. (1890). Principles of economics (8th ed.). Macmillan.
Miller, D. J., & Hayenga, M. L. (2001). Price cycles and asymmetric price transmission in
the US pork market. American Journal of Agricultural Economics, 83(3), 551-562.
174 STUDIA POLITICÆ Nº 64 primavera–verano 2025
Taylor, J. B. (2000). Low ination, pass-through, and the pricing power of rms. European
Economic Review, 44(7), 1389-1408.
