Energy Economics 36 (2013) 658–665

Contents lists available at SciVerse ScienceDirect

Energy Economics
journal homepage: www.elsevier.com/locate/eneco

Volatility spillover between oil and agricultural commodity markets

Saban Nazlioglu a, 1, Cumhur Erdem b, 2, Ugur Soytas c,⁎
a
Department of Econometrics, Pamukkale University, Denizli, Turkey
b
Department of Economics, Gazi Osmanpasa University, Tokat, Turkey
c
Department of Business Administration, Middle East Technical University, Ankara, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: This study examines volatility transmission between oil and selected agricultural commodity prices (wheat,
Received 13 February 2012 corn, soybeans, and sugar). We apply the newly developed causality in variance test and impulse response
Received in revised form 8 November 2012 functions to daily data from 01 January 1986 to 21 March 2011. In order to identify the impact of the food
Accepted 9 November 2012
price crisis, the data are divided into two sub-periods: the pre-crisis period (01 January 1986 to 31 December
Available online 17 November 2012
2005) and the post-crisis period (01 January 2006–21 March 2011). The variance causality test shows that
JEL classification:
while there is no risk transmission between oil and agricultural commodity markets in the pre-crisis period,
O13 oil market volatility spills on the agricultural markets —with the exception of sugar —in the post-crisis period.
C32 The impulse response analysis also indicates that a shock to oil price volatility is transmitted to agricultural
C58 markets only in the post-crisis period. This paper thereby shows that the dynamics of volatility transmission
changes significantly following the food price crisis. After the crisis, risk transmission emerges as another
Keywords: dimension of the dynamic interrelationships between energy and agricultural markets.
Oil prices © 2012 Elsevier B.V. All rights reserved.
Agricultural commodity prices
Volatility spillover

1. Introduction commodities and for some countries commodity exports is an essential

The interest in commodity prices is not a recent phenomenon. How-
ever, the global financial crisis and large fluctuations in commodity
prices have renewed interest on the dynamic relationship between
them. The increasing trend in commodity prices over the last
10 years, which is sometimes attributed to the leading role of the up-
ward trend in crude oil, was followed by a sharp decline when the glob-
al financial crisis hit (Cevik and Sedik, 2011). Spot price of WTI crude oil
increased from $25.56 per barrel in January 1986 to over $138.68 in July
2008. Then following a downward trend it declined down to as low as
$38.95 per barrel by the end of December 2008, before resuming anoth-
er upward trend and reaching $102.36 per barrel on 21 March, 2011, the
last day in our dataset. The wide fluctuations in the price of this impor-
tant commodity raised some questions. One such question is whether
the fluctuations in oil market lead to similar behavior in other commod-
ity markets. And if so, what is the structure of this link? The answers to
these questions are important for investors, traders and policy makers.
Cashin and Pattillo (2000) and Cashin and McDermott (2001)
point out the importance of empirical behavior of commodity prices
and emphasize the relevancy of commodity price cycles for policy pur-
poses. They explain that about 25% of world merchandise trade is on
⁎ Corresponding author. Tel.: +90 312 210 2048; fax: +90 312 210 7962.
E-mail addresses: snazlioglu@pau.edu.tr (S. Nazlioglu), cumhur_erdem@yahoo.com
(C. Erdem), soytas@metu.edu.tr (U. Soytas).
1
Tel.: +90 258 296 2742; fax: +90 258 296 2626.
2 3
Tel.: +90 356 252 2361; fax: +90 356 252 1673.
source of earnings. Therefore, not only commodity price shocks, but also
varying degrees of fluctuations pose serious policy challenges. Cashin
and McDermott (2001) find that over 140 years there has been a decline
in real commodity prices by 1.3% without much evidence of a break in
this downward trend, but a rise in the volatility. They argue that policy
implications of the increase in volatility are more important than con-
cerns about the long-run downward trend. This is because sharp move-
ments in commodity prices have serious impacts on terms of trade, real
incomes and fiscal positions of countries that depend on commodities.
The adverse effects of increasing world oil prices and possible mit-
igation policies have been the center of interest for manufacturer and
consumer group levels in many sectors as well as governments of oil
dependent economies. Like other sectors, agricultural commodity
markets are also affected by fluctuations in oil prices at various levels.
Especially with the introduction of biofuels, effects of oil markets on
agricultural commodity markets are believed to intensify. Increase in
the number and intensity of biofuel mandates also contributes to the
link between agricultural and energy prices. It can be argued that bio-
fuel policies linking agricultural and energy markets led to the food
price crisis in 2006. 3 Furthermore, financialization of the commodity
markets seems to have increased the degree of integration between
energy and agricultural commodity markets.
Bioethanol (primarily from corn) and biodiesel (primarily from soy-
beans) are considered to be technological substitutes for conventional
We thank an anonymous referee for pointing out this issue.

0140-9883/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.eneco.2012.11.009
 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665
fuels such as diesel and gasoline (Chang and Su, 2010). Since production
of these biofuels is heavily dependent on supply of agricultural com-
modities, it is expected that there is a tight market integration between
energy and agricultural markets (Schmidhuber, 2007), and this market
integration is likely to be the most important change to occur in agricul-
ture in decades (Tyner and Taheripour, 2008). As production of biofuels
continues to rise, demand for these commodities is expected to increase
which causes food prices to increase. Both crude oil and agricultural
commodity prices displayed exceptional volatility throughout most of
2008 (Du et al., 2011). A high and positive price correlation between
crude oil and corn at 0.80 for the period of 2006–2008 was observed
(Tyner, 2008). Mitchell (2008) stated that one of the most important
factors that cause food prices to increase in the 2006–2008 period is
the large increase in US and EU biofuels production. He also added
that around 70% of the increase in international food prices was attrib-
uted to biofuels and the related consequences of low grain stocks, large
land use shifts, speculative activity, and export bans. Yang et al. (2008)
stated that respectively, 38% and 22% of the increases in the interna-
tional corn and wheat prices can be attributed to biofuels in the
2005–2008 period.
The rising food prices during the recent years have raised the ques-
tion of whether oil markets have any explanatory power on the recent
upward movements in agricultural food prices. The food-energy nexus
has become a controversial issue. Many researchers indicate that in-
creasing oil prices is the main factor behind the recent major demand
shock experienced by agricultural markets (e.g. Abbott et al., 2008;
Baffes, 2007; Chang and Su, 2010; Collins, 2008; Mitchell, 2008;
Rosegrant et al., 2008; Yang et al., 2008). In contrast, some researches
indicate that there is no direct relationship between oil and agricul-
tural commodity prices. For example, Zhang et al. (2010) argued that
oil price increases do not have direct impacts on agricultural commod-
ity prices. Pindyck and Rotemberg (1990) examined the co-movement
of wheat, cotton, copper, gold, crude oil, lumber and cocoa prices and
found that the cross-price elasticities of demand and supply are zero.
Gilbert (2010) stated that there is no direct causal relationship be-
tween oil and agricultural prices and the correlation between oil and
agricultural prices is due to demand growth and monetary and finan-
cial developments. His findings do not offer support for restrictions on
the use of food commodities as biofuel production. According to the
Council of Economic Advisors only around 3% retail food price increase
can be attributed to the ethanol production in 2007 (Lazear, 2008).
In addition to the increase in production of biofuels, another factor
that boosts agricultural prices is the active interests of global inves-
tors, traders, and speculators in agricultural commodity markets. Re-
cently, players in global financial markets view commodity markets
as alternative investment areas for hedging and portfolio diversifica-
tion purposes (Baffes and Haniotis, 2010; Sari et al., 2011). In other
words, agricultural commodities are being regarded as financial as-
sets (Alom et al., 2011). Therefore, intuitively, commodity markets
may be subject to similar dynamics as financial markets. Globalization
and increased integration of world markets have accelerated the
“financialization of commodities.” Apparently, sound investment de-
cisions and policy options call for a better understanding of the recent
dynamics of commodity markets. Therefore, global investors and gov-
ernments need to analyze the dynamics of price and risk transmis-
sions among commodity markets. “Financialization of commodities”
has added a new dimension to the determinants of commodity price
hikes (Baffes and Haniotis, 2010; Sari et al., 2011). The financial conta-
gion literature notes the increased correlation across international
financial markets during a crisis. A similar phenomenon is observed
between commodity markets. Baffes and Haniotis (2010) point out
the expectation of a stronger link between energy and non-energy
commodity prices, especially food markets.
The literature on energy and agricultural market linkages shows
that price transmission between oil and agricultural commodity prices
is extensively analyzed for different time frameworks, using different
659
econometric techniques, and for various countries. 4 However, as
indicated above, the relationship between oil and agricultural com-
modity prices is complicated and suggests the need to take energy-
agricultural markets integration into account. In that respect, the re-
cent tendency in the literature on the energy-food nexus is to focus
on the volatility spillover between oil and agricultural commodities.
It is now well known that energy and agricultural markets are recently
characterized by more volatile dynamics that call for deeper analyses
of volatility transmission between these markets.
The goal of this paper is to identify whether volatility in oil prices
have any explanatory impact on the volatility in agricultural commod-
ity prices. We concentrate on identifying the structure of the volatility
transmission mechanism between oil and the selected agricultural
commodities (wheat, corn, soybeans, and sugar) that are the main
crops used as inputs in production of biofuels and are the key agricul-
tural products for food in the world. The newly developed causality in
variance test and impulse response functions are applied to the data
covering the period 01 January 1986–21 March 2011. To investigate
the impact of the 2006–2008 food price crisis on volatility transmis-
sion mechanism between energy and agricultural markets, the empir-
ical analysis is conducted for two sub-periods (the pre-crisis period
from 01 January 1986 to 31 December 2005, and the post-crisis period
01 January 2006–21 March 2011). The empirical results provide evi-
dence on volatility spillover from oil prices to agricultural commodity
prices in the post-crisis period, implying that agricultural commodity
markets have become more integrated with energy markets after the
crisis.
A considerable body of research exists on the linkages between
crude oil and agricultural commodity prices. Our study differs from
other studies by employing a newly developed causality in variance
test. To the best of our knowledge, within the energy-food nexus
this is the first study to carry out the causality in variance test devel-
oped by Hafner and Herwartz (2006) which is superior to test devel-
oped by Cheung and Ng (1996). This paper also utilizes impulse
response functions in order to identify temporal volatility transmis-
sion dynamics between energy and agricultural markets.
The paper is organized as follows: The literature is reviewed in
Section 2. Econometrics methods are outlined in Section 3, followed
by data definition in Section 4. Interpretation of and discussion on
empirical results are droved in Section 5. Finally, Section 6 is devoted
to a brief summary of this paper and policy implications.
2. Literature review
There is a vast literature on the link between oil and agricultural
prices but the nature of this causal link remains unclear. There is an
increased interest in volatility spillover or risk transfer between oil
and agricultural commodity prices. However, this literature seems
to be still scarce and calls for more attention to the dynamics of the
risk transmission. In the literature, a distinction can be made among
three key linkages between oil and agricultural prices: (i) oil as a pro-
duction cost; (ii) biofuels; and (iii) co-movement with agricultural
commodities due to investment fund activity. 5 Table 1 presents a
chronological summary of the literature in terms of method types,
commodity, data and key findings.
The relationship between oil and agricultural prices in terms of oil
as a production cost in agriculture was examined by Baffes (2007,
2010), Harri and Hudson (2009), Chang and Su (2010), Alghalith
(2010), Du et al. (2011), Alom et al. (2011). Baffes (2007) analyzed
how crude oil prices spill on the prices of 35 internationally traded
primary commodities and found that the pass-through of crude oil
4
We refer interested readers to Nazlioglu and Soytas (2011), Nazlioglu (2011),
Nazlioglu and Soytas (2012) for the literature reviews on the studies that examine
price transmission between oil and agricultural prices.
5
We thank an anonymous referee for suggesting this classification.
660 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665

Table 1
Summary of literature review.

Study Data Method Commodity Key findings

Baffes (2007) 1960–2005 (annual) OLS Crude oil 35 commodities Pass-through of crude oil price changes to the overall non-energy
commodity index, the fertilizer index, agriculture and metals
Harri & Hudson (2009) 2003–2009 (daily) Causality Crude oil, corn, soybeans Volatility spillover from oil prices to corn prices after the food price crisis
Krichene (2008) 2000–2007 (daily) LPGHT Crude oil Expansionary monetary policies during the early 2000s affect oil prices
Alghalith (2010) 1974–2007 (annual) NOLS Crude oil, food basket An increase in oil price and its volatility yields a higher food price
Baffes (2010) 1960–200 (annual) OLS Energy index Non-energy index The fertilizer index exhibits the largest pass-through followed by
agriculture and metals.
Busse et al. (2010) 2002–2009 (weekly) MS-VECM Biodiesel, rapeseed oil, soya oil, Strong impact of crude oil price on biodiesel prices, and of biodiesel
crude oil prices on rapeseed oil prices
Chang and Su (2010) 2000–2008 (daily) EGARCH Crude oil, corn, soybean Volatility spillover from crude oil to corn and soybean prices
Zhang et al. (2010) 1989–2008 (monthly) VECM Ethanol, corn, rice, soybeans, No direct long-run price relations between fuel and agricultural
sugar, wheat, gasoline, crude oil commodity prices
Alom et al. (2011) 1995–2010 (daily) VAR Crude oil, food price index Positive correlation between indices of world oil and food producers
price volatility. The results vary across countries and sub-periods.
Cevik and Sedik (2011). 1990–2010 (monthly) OLS Crude oil, fine wine Macroeconomic factors are the main determinants of commodity prices
Du et al. (2011) 1998–2009 (weekly) SVM Crude oil, corn, wheat Volatility spillover among crude oil, corn, and wheat markets
Serra (2011) 2000–2009 (monthly) SP-GARCH Ethanol, crude oil, sugar Shocks on crude oil and sugar market cause an increase in the volatility
of ethanol price.
Serra et al. (2011) 1990–2008 (monthly) ST-VECM Ethanol, corn, crude oil, Long-run relationships among the prices as well as strong links between
gasoline energy and food prices
Kristoufek et al. (2012) 2003–2011 MS&HT Biodiesel, ethanol, fuels, corn, Food and fuel prices affect biofuel prices and biofuel prices has a limited
(weekly and monthly) wheat, soybean, sugar cane capacity to determine food prices
and beets.
Hassouneh et al. (2012) 2006–2010 (weekly) MLLR Biodiesel, sunflower, crude oil Long-run equilibrium relationship among the three prices

OLS: ordinary least squares, NOLS: non-linear OLS, SVM: stochastic volatility models, SP-GARCH: Semi-parametric GARCH model, LPGHT: Levy process of generalized hyperbolic
type, MS&HT: minimal spanning and hierarchical trees, MLLR: multivariate local linear regression, ST-VECM: smooth transition vector error correction model, MS-VECM:
Markov-switching vector error correction model.

price changes to the overall non-energy commodity index, the fertil-
izer index, agriculture and metals are 0.16, 0.33, 0.17 and 0.11, respec-
tively. At a more disaggregated level, Baffes (2010) re-examined this
relationship and concluded that the highest pass-through of oil price
changes is to the fertilizer index followed by agriculture.
The link between oil and agricultural prices in terms of biofuels
was examined by Serra (2011), Serra et al. (2011), Hassouneh et al.
(2012), Kristoufek et al. (2012), Busse et al. (2010) and Zhang et al.
(2010). Serra (2011) analyzed the volatility spillover between crude
oil, ethanol and sugar prices in Brazil. She found that there are strong
volatility relationships between the prices and that crude oil and sugar
market shocks cause an increase in the volatility of the ethanol price.
This proves the existence of a dynamic link between biofuel, fuel,
and agricultural markets. In a different study, Serra et al. (2011) ana-
lyzed the price linkages and transmission patterns in the U.S. ethanol
industry and found that there exist long-run relationships among
the prices of ethanol, corn, oil and gasoline as well as strong links be-
tween energy and food prices. Similar to Serra (2011), they uncover a
long run equilibrium relationship between oil, biofuel and agricultural
markets. Hassouneh et al. (2012) examine the price linkages and price
transmission patterns between food and energy prices in Spain. They
found out that there is a long-run, equilibrium relationship between
biodiesel, sunflower and crude oil prices; that biodiesel is the only var-
iable that adjusts to deviations from the long-run relationship and that
sunflower oil prices are influenced by energy prices through short-run
price dynamics. Kristoufek et al. (2012) analyze the existence of the
relationship between biodiesel, ethanol and related fuels and com-
modity prices in the US and Germany. Their results show that even
though biofuel is affected by food and fuel prices, biofuel prices has a
limited capacity in the determination of food prices. They also found
that the relationship between prices changes depending on the data
frequency used. Busse et al. (2010) investigated the vertical price
transmission in the biodiesel supply chain in Germany by focusing
on the connections between prices of rapeseed oil, soy oil, biodiesel
and crude oil. They found evidence for a strong impact of crude oil
price on biodiesel prices, and of biodiesel prices on rapeseed oil prices.
Zhang et al. (2010) analyze both short- and long-run relationship
between prices of fuel and agricultural commodities. Their results
show that there is no direct long-run price relation between fuel and
agricultural commodity prices and there is only limited, if any, direct
short-run relationships.
With respect to the literature in terms of co-movement of com-
modity prices with macroeconomic factors and financial indicators,
Krichene (2008) investigates oil price movements between January
2000 and October 2007. He argues that the rapid increase recently
observed in oil and other commodity prices can be attributed to the
expansionary monetary policies during the early 2000s. Due to the im-
pact of expansionary policies, world demand for commodities rose
while supply lagged behind, putting upward pressure on all commod-
ity prices. He points out the new dynamics in commodity prices due to
easing of monetary stance in the late 2007 and early 2008. Du et al.
(2011) investigated the role of speculation in driving crude oil price
variation after controlling for other influencing factors. They also
attempt to quantify the extent to which volatility in the crude oil mar-
ket pass through into agricultural commodity markets (corn and
wheat) in the U.S. They found evidence of volatility spillover among
crude oil, corn, and wheat markets after the fall in 2006, implying
risk transfer between these commodity markets. Cevik and Sedik
(2011) examine the extreme fluctuations in commodity prices (oil
and wine) between 1990 and 2010. According to their results, macro-
economic factors emerge as the main determinants of commodity
prices. They also point out that, although major economies account
for approximately half of world consumption, emerging markets
make up a significant incremental addition to demand. Whatever the
reason behind the long run trend or short run fluctuations is, not
just crude oil but other commodity spot prices followed a similar pat-
tern. They all experienced steadily increasing trends first and then an
unprecedented slump after the global crisis. Reviewing several alter-
native theories of the crude oil price hike in 2008, ranging from
demand and supply dynamics to commodity speculation, Hamilton
(2009) concludes that instead of being alternative explanations, they
may jointly be responsible for the price shock.
As regards to risk transfer between energy and agricultural mar-
kets, in one of the early attempts to discover volatility spillover,
 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665 661

Harri and Hudson (2009) found that there is volatility spillover from hypothesis of non-causality in variance between two return series is
oil futures prices to corn futures prices after the food price crisis. In described as follows:
line with this finding, Chang and Su (2010) provided evidence of vol- 

atility spillover from crude oil to corn and indicted that soybean prices
ðj Þ
H 0 : Var εit
F t−1 ¼ Varðεit jF t−1 Þ j ¼ 1; …; N; i≠j ð1Þ
are positively significant during the higher crude oil price period,
implying an economic substitution effect during the higher crude oil
where Ft(j) = Ft \ σ(εjτ, τ ≤t) and εit are the residuals from GARCH model.
price period.
The following model is considered to test for the null hypothesis.
In addition to volatility transmission between world oil and world
agricultural commodities, some studies concentrated on the volatility qffiffiffiffiffiffiffiffiffiffiffi  ′
2 2
transmission from world oil prices to domestic agricultural commod- εit ¼ ξit σ 2it g t ; g t ¼ 1 þ z′jt π; zjt ¼ εjt−1 ; σ jt−1 ð2Þ
ity prices. For instance, Alghalith (2010) analyzed the impact of oil
price uncertainties on food prices in Trinidad and Tobago and found where conditional variance σit2 = ωi + αiεi,t−1 2 2
+ βiσi,t−1 and ξit denotes
that an increase in oil price and its volatility yields a higher food the standardized residuals of GARCH model. In equation (2), the suffi-
price. The finding that a higher risk in the oil market leads to a higher cient condition for equation (1) is π = 0 which ensures that the null hy-
food price suggests that there exists a risk transfer mechanism be- pothesis of non causality in variance H0 : π= 0 is tested against the
tween the two commodity markets. Alom et al. (2011) investigated alternative hypothesis H1 : π≠ 0. The score of the Gaussian log-
volatility spillover from world oil prices to food prices for the selected likelihood function of εit is given by xit(ξit2 − 1)/2 where the derivatives
Asia and Pacific countries and found that food price volatility is positive- xit = σit−2(∂ σit2/∂ θi) and θi = (ωi, αi, βi)′. Hafner and Herwartz (2006)
ly correlated with world oil price volatility but the results vary across propose the following LM test in order to test the volatility transmission
countries and sub-periods. Furthermore, their results imply that volatil- between the two series:
ity spillover from world oil to domestics food prices are stronger for
T 
X 
 T 
X 

the more recent sub-period, indicating increasing interdependency be- 1 2 −1 2
tween world oil and Asia Pacific agricultural markets during the recent λLM ¼ ξit −1 z′jt V ðθi Þ ξit −1 zjt ð3Þ
4T t¼1 t¼1
years.
Apparently there is a rapidly emerging literature on the dynamics where
of price transmissions between energy and agricultural commodity
markets. The conclusions appear to be mixed. The differences may !−1 !
κ X T XT XT X
T
be arising from different time periods, data sets, frequencies, method- V ðθi Þ ¼ zjt z′jt − zjt x′it xit x′it xit z′jt ;
4T t¼1
ologies and models utilized. Furthermore, although there is some t¼1 t¼1 t¼1
T  2
evidence on the financialization of commodity markets, there aren't 1X 2
κ¼ ξit −1
many studies that consider risk transmissions across these markets. T t¼1
Additionally, just like the financial market dynamics, the relationship
between commodity markets is subject to change around crises. This The asymptotic distribution of the test statistic in Eq. (3) will de-
paper attempts to fill the gap in the literature regarding the volatility pend on the number of misspecification indicators in zjt. Since there
transmissions between energy and agricultural markets before and are two misspecification indicators in λLM, the test has an asymptotic
after the commodity crisis in 2006. Our results show that the dynamics chi-square distribution with two degrees of freedom.
of volatility spillover changes significantly following the 2006 commod- In addition to testing for causality in variance between world oil
ity crisis. and agricultural commodity prices, this study employs impulse-
response analysis in order to determine how the volatility of agricul-
3. Econometric methodology tural commodity prices respond to a shock in world oil price volatility.
In this regard, we utilize the generalized impulse-response method
In order to assess the volatility spillover between world oil and ag- advocated by Koop et al. (1996), and Pesaran and Shin (1998) which
ricultural commodity prices, this study adopts causality in variance is superior to the traditional approach because it is not subject to the
test recently developed by Hafner and Herwartz (2006). In examining orthogonality critique arising from Cholesky ordering and the results
volatility spillover between two series, Cheung and Ng (1996) and from the generalized impulse response functions are not sensitive to
Hong (2001) developed a causality-in-variance test which is based on the ordering of variables in the vector autoregression (VAR) system.
cross-correlation functions (CCF) of standardized residuals obtained
from univariate general autoregressive conditional heteroscedasticity 4. Data
(GARCH) estimations. The CCF based Portmanteau test is likely to be
suffering from significant oversizing in small and medium samples We employ daily data covering the period 01 January 1986 and 21
when the volatility processes are leptokurtic (Hafner and Herwartz, March 2011 for the spot prices of world oil, corn, soybeans, wheat,
2006: 140). In addition, results from CCF based testing approach are and sugar. The data on all series is compiled from DataStream. 6 The
sensitive to the orders of leads and lags which in turn questions the ro- return series — ln(pt/pt − 1) where pt is the price at day t — are used
bustness of findings. The volatility spillover test of Hafner and Herwartz in the empirical analysis and the natural logarithms of the variables
(2006) based on Lagrange multiplier (LM) principle overcomes the are arranged in 5-day weeks. Regarding the oil price-agricultural
shortfalls of Cheung and Ng's method and is very practical for empirical commodity prices nexus, it has now been argued that agricultural
illustrations. Furthermore, the Monte Carlo experiment carried out in commodity prices are not responsive to the oil prices until 2006
Hafner and Herwartz (2006) indicates that the LM approach is more ro- (Campiche et al., 2007). The food price crisis during the 2006–2008
bust against leptokurtic innovations in small samples and the gain from period results in higher correlation between the oil and agricultural
carrying the LM test increases with sample size. The results further commodity prices. To account for the impact of the food price crisis,
show that an inappropriate lead and lad order choice in the CCF test dis- it is convenient to divide the full sample into subgroups (for example,
torts its performance and thereby leads to the risk of selecting a wrong
order of the CCF statistic. In what follows, we briefly explain the details
of Hafner and Herwartz (2006) causality in variance test. 6
Even though daily world spot prices for coffee, cocoa, rice, and sunflower seed are
In the Hafner and Herwartz (2006) approach, testing for causality available at DataStream, we did not use these series in the empirical analysis because
in variance is based on estimating univariate GARCH models. The null of lack of data variability over the period employed.
662 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665

Campiche et al., 2007; Harri and Hudson, 2009; Harri et al., 2009; Table 3
Nazlioglu, 2011; Nazlioglu and Soytas, 2011). We therefore work Correlation matrix.

with two subgroups: (i) the pre-crisis period spanning from 01 Janu- Panel A:
ary 1986 to 31 December 2005, and (ii) the post-crisis period cover-
Pre-crisis period Oil Corn Soybeans Wheat Sugar
ing the time framework from 01 January 2006 to 21 March 2011.
Oil 1
Table 2 represents the descriptive statistics for the sub-periods. It
Corn 0.042 1
seems that data characteristics of the return series are slightly different Soybeans 0.039 0.571 1
in the pre- and post-crisis periods. First of all, as expected, the mean and Wheat 0.039 0.392 0.325 1
the volatility of the returns in the post-crisis period are higher than Sugar −0.003 0.072 0.061 0.0645 1
those in the pre-crisis period. Secondly, the mean, standard deviation,
Panel B:
skewness and kurtosis of the world oil returns are greater than those
of the agricultural commodity returns in the pre-crisis period. On the Post-crisis period Oil Corn Soybeans Wheat Sugar
other hand, the agricultural commodity returns have higher mean and Oil 1
skewness than the oil returns. In the post-crisis period, standard devia- Corn 0.337 1
Soybeans 0.398 0.608 1
tion for corn and soybeans are substantially higher than those of the
Wheat 0.273 0.533 0.439 1
other variables. This in fact is expected due to the fact that the oil Sugar 0.241 0.233 0.221 0.181 1
price surge during the recent years increases the derived demand for
Correlation coefficient series are for log return series.
the agricultural commodities such as corn and soybeans which are
used in biofuels production which in turn leads to a substantial rise in
the prices of those commodities. It seems that all estimated coefficients are statistically meaningful
The different data characteristics in the pre- and post-crisis periods at 1% level of significance. The coefficients in the variance equations
raise the question of whether the correlation between the world oil are positive and show that the conditional variance process of the
and agricultural commodity returns vary across the sub-periods. In commodity returns are convergent. The volatility processes of oil,
this respect, Table 3 illustrates the correlation matrix among the series corn, soybeans, and wheat returns do not change drastically from
in question. It is clear that the correlation between the oil and agricul- one period to the other. More specifically, the ARCH parameter (α)
tural commodity returns dramatically increased in the post-crisis pe- and the GARCH parameter (β) appear to be similar across the two
riod compared to the pre-crisis period. Another interesting result is sub-samples. The degree of persistence (α + β) in each period shows
that even though the correlation between corn, soybeans, and wheat that the volatility shocks are very persistent which implies a fair
returns increased in the post-crisis period to some extent, the correla- amount of persistence of volatility shocks on the daily returns. The
tion between sugar and other commodity returns as well as the oil sugar returns have a different volatility path than other commodities
returns are significantly higher in the post-crisis period than the between the sub-periods. We observe that the ARCH and GARCH pa-
pre-crisis period. The descriptive analysis indicates that the oil and ag- rameters in the pre-crisis period are considerably higher than those
ricultural commodity prices are likely to be determined by the new in the pre-crisis period. Although, the degree of the persistence of
price regimes during the recent years and that they seem to be signif- the sugar returns show higher persistency in both sup-periods, the
icantly responsive to the variations in each other. shock seems to be very persistent in the pre-crisis period in compari-
son to the post-crisis period.
Large values of the ARCH and GARCH parameters influence the
5. Empirical findings conditional volatility in different ways. A high ARCH parameter

In order to examine the volatility spillover between oil and agricul- Table 4
tural commodity returns, we first estimate the univariate GARCH(1,1) Results for variance equations.
processes. Table 4 reports the results for variance equations of the Pre-crisis period Post-crisis period
GARCH model estimations for both the pre- and post-crisis periods. 01 January 1986–31 01 January 2006–31
We first check whether the stability conditions of the GARCH model December 2005 March 2011
hold which impose the constraints ω > 0, 0 ≤ α, 0 ≤ β, α + β b 1. All of Oil
the estimated GARCH models satisfy the stability conditions. We ω 5.72E − 06 [0.0000] 1.05E − 05 [0.0006]
therefore proceed to draw some inferences. α 0.096437 [0.0000] 0.076378 [0.0000]
β 0.902193 [0.0000] 0.905506 [0.0000]

Table 2 Corn
Descriptive statistics. ω 4.76E − 06 [0.0000] 1.52E − 05 [0.0002]
α 0.078802 [0.0000] 0.050617 [0.0000]
Panel A: β 0.902800 [0.0000] 0.922778 [0.0000]
Pre-crisis Oil Corn Soybeans Wheat Sugar
Soybeans
perioda
ω 2.34E − 06 [0.0000] 3.58E − 06 [0.0014]
Mean 1.67E − 04 −3.37E − 05 2.25E − 05 −1.63E − 05 2.33E − 04 α 0.082600 [0.0000] 0.048096 [0.0000]
Std. Dev. 0.025 0.015 0.014 0.018 0.021 β 0.909353 [0.0000] 0.942405 [0.0000]
Skewness −1.047 −0.406 −0.839 −0.439 −0.294
Kurtosis 21.386 8.299 11.760 21.906 10.262 Wheat
ω 1.23E − 05 [0.0000] 1.83E − 05 [0.0003]
Panel B: α 0.091041 [0.0000] 0.067462 [0.0000]
β 0.879825 [0.0000] 0.915860 [0.0000]
Post-crisis periodb

Mean 3.80E − 04 8.75E − 04 6.00E − 04 5.78E − 04 5.19E − 04 Sugar

Std. Dev. 0.026 0.023 0.019 0.031 0.022 ω 2.63E − 06 [0.0000] 9.03E − 05 [0.0012]
Skewness 0.096 −0.200 −0.624 −0.461 −0.181 α 0.024599 [0.0000] 0.089749 [0.0000]
Kurtosis 8.085 5.188 6.443 6.599 4.512 β 0.969841 [0.0000] 0.725776 [0.0000]
a
01 January 1986–31 December 2005. Variance equation: σt2 = ω + αεt−1
2 2
+ βσt−1. The results are obtained from GARCH(1,1)
b
01 January 2006–21 March 2011. estimations. Numbers in brackets are p-values.
 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665 663

Table 5 find, on the one hand, unidirectional volatility spillover from the oil
Result for causality in variance test. market to the corn market. On the other hand, there is also evidence
Pre-crisis period Post-crisis period of a bidirectional volatility spillover between oil-soybeans and oil-
01 January 1986–31 01 January 2006–31 wheat markets. However, similar to the pre-crisis period, there is no
December 2005 March 2011 volatility spillover mechanism between oil and sugar returns.
Oil ≠ > corn 4.141 [0.1261] 9.939 [0.0069] Even though volatility spillover between oil and agricultural com-
Corn ≠ > oil 0.064 [0.9685] 4.024 [0.1337] modity markets during the post-crisis period was expected, it is
Oil ≠ > soybeans 0.923 [0.6304] 8.809 [0.0122]
somewhat interesting to find out Granger causality in variance run-
Soybeans ≠ > oil 0.031 [0.9847] 5.943 [0.0512]
Oil ≠ > wheat 1.146 [0.5638] 10.455[0.0054] ning from wheat to oil returns in the pre-crisis period. However,
Wheat ≠ > oil 12.103 [0.0024] 5.454 [0.0654] there may be plausible explanations for this finding. First of all, as
Oil ≠ > sugar 2.843 [0.2413] 0.601 [0.7405] pointed out by Pindyck and Rotemberg (1990), traders in different
Sugar ≠ > oil 1.647 [0.4389] 2.894 [0.2353] commodity markets may be responding similarly to noneconomic
Numbers in brackets are p-values. factors. This tandem behavior may be driving the Granger causality
from one market to the other. Second, the recent surge in the oil
prices has been forcing the developed oil importing economies such
implies that the effects of a shock are more pronounced in the subse- as the US, and EU countries to lend more support for the production
quent period. In contrast, a high GARCH parameter implies that the of biofuels. This relatively recent tendency makes agricultural com-
effects of a shock are more persistent (Enders, 2004: 134). Therefore, modities which are used in biofuels production such as corn, soy-
while a high ARCH parameter implies high short-run volatility, high beans, and sugar cane more important than they were in the past.
GARCH parameters indicate high long-run volatility. The results Third, wheat is the major food product in all over the world which
from the GARCH model estimations clearly show that the volatility leads international investors to concentrate on the price dynamics
processes of the commodity returns in question is dominated by the of that product in order to develop the investment strategies and
GARCH effect. This finding implies that the conditional variance dis- portfolio diversification.
play more autoregressive persistence which suggest high long-run In the final step of the empirical modeling strategy we define how
volatility in the commodity returns. and to what extent the volatility of the agricultural commodity
After determining the volatility processes of the commodity returns respond to a shock in oil volatility in the short-run. In this re-
returns, we now concentrate on investigating whether there are gard, the generalized impulse-response functions are derived from
volatility spillovers between the oil and the agricultural commodity the VAR models for both the pre- and post-crisis periods. The optimal
returns. To this end, the Hafner and Herwartz (2006) causality in var- lag lengths in the VAR system are determined via the Schwarz infor-
iance test is carried out and the results are illustrated in Table 5. The mation criterion. Since all roots of the VAR model are within the unit
results for the pre-crisis period show that there is no volatility spill- circle, VAR system satisfies the stability condition.
over from the oil returns to the agricultural commodity returns. This The impulse response functions for one standard deviation shock
is also true for the volatility spillover from the agricultural commodity to the oil returns volatility are presented in Figs. 1 and 2 for the
returns to the oil returns with the exception of wheat. The findings pre-crisis period and the post-crisis period, respectively. The results
show that there is a volatility spillover from the wheat to the oil indicate that none of the impulse response functions are significant.
returns in the pre-crisis period. Hence, a shock in the oil market volatility is not transmitted to the ag-
The results for the volatility spillover relationships between the oil ricultural commodity market volatility in the short-run.
and agricultural commodity returns are slightly different in the post- The volatility transmission from the oil returns to the agricultural
crisis period than the pre-crisis period. In the post-crisis period, we returns turns out to be significant in the post-crisis period. It appears

Response of corn to oil Response of soybeans to oil

.000004 .000004

.000002 .000002

.000000 .000000

-.000002 -.000002

-.000004 -.000004

-.000006 -.000006

-.000008 -.000008
25 50 75 100 125 150 25 50 75 100 125 150

Response of wheat to oil Response of sugar to oil

.000008 .000010

.000004 .000008

.000006
.000000
.000004
-.000004
.000002
-.000008 .000000

-.000012 -.000002
25 50 75 100 125 150 25 50 75 100 125 150

Fig. 1. Generalized impulse-response functions for the pre-crisis period.

664 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665
Response of corn to oil
.000020
.000016
.000012
.000008
.000004
.000000
-.000004
25 50 75 100 125
Response of wheat to oil
.00005
.00004
.00003
.00002
.00001
.00000
-.00001
25 50 75 100 125
Fig. 2. Generalized impulse-response functions for the post-crisis period.
that the behavior of the volatility of the corn and soybeans returns
with respect to a shock in the volatility of the oil returns have similar
patterns. In particular, a volatility shock in oil returns causes a posi-
tive immediate response of the volatility of corn and soybeans returns
within the week of the shock. This positive impact seems to be con-
tinuing over three weeks. After that, the magnitude of the positive
impact starts to decrease and eventually becomes insignificant within
approximately 60 days. Although the dynamics of the volatility in
wheat market resembles those of corn and soybeans, it has some dif-
ferences. Specifically, the volatility shock in oil returns increases the
volatility of wheat throughout two weeks; it lasts only one month be-
cause the impulse response functions are insignificant after approxi-
mately 30 days. With respect to the response of the sugar volatility
to a shock in the oil volatility, the impulse response functions clearly
show that the immediate positive impact dies out within 10 days.
Overall, impulse responses show that risk transmission is more rele-
vant following the crisis.
6. Conclusions and implications
This paper investigates volatility spillover between oil and selected
agricultural commodity markets (wheat, corn, soybean, and sugar)
that are key agricultural products for biofuels and for food in the
world. The data spanning from January 01, 1986 to March 21, 2011
are divided into two sub-periods as January 01, 1986 to December 31,
2005 and January 01, 2006 to March 21, 2011 (the post-crisis period)
to account for the 2006–2008 commodity crisis. In the descriptive anal-
ysis, we observed that commodity market returns follow similar prop-
erties as financial returns. They are non-normal with excess kurtosis
and skewness and the error terms show significant ARCH effects.
Using a relatively new test for causality in variance by Hafner and
Herwartz (2006), we find out that there is no risk transfer between
any other markets before the commodity crisis; however, it is apparent
that after the crisis oil market risk is transmitted to corn, wheat, and
soybean markets. Moreover, we discover volatility spillover from
wheat to oil market for both periods. The sugar market on the other
hand seems to be neutral to oil market risks. The volatility responses
of all agricultural markets to risk shocks in oil markets seem to be signif-
icant only for the post-crisis period. These responses are not permanent
and die off in about a month. Therefore, our results strongly support
Response of soybeans to oil
.000020
.000016
.000012
.000008
.000004
.000000
-.000004
150 25 50 75 100 125 150
Response of sugar to oil
.000016
.000012
.000008
.000004
.000000
-.000004
150 25 50 75 100 125 150
that interdependency between energy and agricultural markets has in-
creased during the recent years, which is consistent with the results
obtained by Harri and Hudson (2009), Chang and Su (2010) and Serra
et al. (2011).
It is well established that oil markets have always attracted global
investors. The fact that significant risk transmissions are observed
from oil to agricultural commodity markets after the crisis suggests
that investors seeking safe havens increase the financialization of the
agricultural commodity markets. This suggests that local measures
to suppress price uncertainty in agricultural markets may not be effec-
tive in the short run. The global factors such as the risk in energy mar-
kets seem to drive the short run volatility in agricultural markets. The
importance of this effect is higher for countries that are most vulnera-
ble to food price fluctuations. More light can be shed on this issue by
studying the responses of local agricultural prices to global energy
shocks. Furthermore, investors interested in commodity markets
should realize that risk in one commodity market may not be indepen-
dent of risk in other commodity markets. Further investigation is nec-
essary in order to find out how risk transmission mechanism works
between different financial and commodity markets.
The dynamics of commodity prices are complicated and different
factors may be affecting these markets. The recent discussion suggests
that in addition to the energy-agriculture linkage, the financial factors
such as exchange rates, futures markets, speculation, and interest
rates may play role on the recent dynamics of commodity prices.
Thereby, investigating agricultural price determination suggests the
need for more empirical analysis to jointly identify the impacts of var-
ious factors on agricultural prices. In that respect, future studies can
extend the literature at least in two ways. First, researches can benefit
from the flexibility of structural vector autoregression models to im-
pose theoretical restrictions to identify impacts of energy and finance
markets on agricultural markets. Secondly, multivariate volatility
spillover analysis which provides room to apply causality in variance
tests and impulse response analysis based on multivariate models
may provide new insights.
References
Abbott, P.C., Hurt, C., Tyner, W.E., 2008. What's driving food prices? Farm Foundation
Issue Report, July 2008.
 S. Nazlioglu et al. / Energy Economics 36 (2013) 658–665
Alghalith, M., 2010. The interaction between food prices and oil prices. Energy Econ. 32,
1520–1522.
Alom, F., Ward, B., Baiding, H., 2011. Spillover effects of world oil prices on food prices:
Evidence from Asia and Pacific countries. 52nd New Zealand Association of Econo-
mists Annual Conference, Wellington, New Zealand, 29 June–1 July 2011.
Baffes, J., 2007. Oil spills on other commodities. Resour. Policy 32, 126–134.
Baffes, J., 2010. More on the energy/nonenergy price link. Appl. Econ. Lett. 17 (16),
1555–1558.
Baffes, J., Haniotis, T., 2010. Placing the 2006/08 commodity price boom into perspec-
tive. Policy Research Working Paper 5371. World Bank.
Busse, S., Brummer, B., Ihle, R., 2010. Interdependencies between fossil fuel and renew-
able energy markets: The German biodiesel market. DARE Discussion Paper 1010.
Campiche, J.L., Bryant, H.L., Richardson, J.W., Outlaw, J.L., 2007. Examining the evolving
correspondence between petroleum prices and agricultural commodity prices. The
American Agricultural Economics Association Annual Meeting, Portland, OR, July
29-August 1, 2007.
Cashin, P., McDermott, C.J., 2001. The Long-Run Behavior of Commodity Prices: Small
Trends and Big Variability. Working Paper 01/68. International Monetary Fund,
Washington.
Cashin, P., Pattillo, C., 2000. Terms of Trade Shocks in Africa: Are They Short-Lived or
Long-Lived? Working Paper 00/72. International Monetary Fund, Washington.
Cevik, S., Sedik, T.S., 2011. A Barrel of Oil or a Bottle of Wine: How Do Global Growth
Dynamics Affect Commodity Prices? Working Paper 11/01. International Monetary
Fund, Washington.
Chang, T., Su, H., 2010. The substitutive effect of biofuels on fossil fuels in the lower and
higher crude oil price periods. Energy 35 (7), 2807–2813.
Cheung, Y.W., Ng, L.K., 1996. A causality in variance test and its application to financial
market prices. J. Econ. 72, 33–48.
Collins, K., 2008. The role of biofuels and other factors in increasing farm and food
prices: a review of recent developments with a focus on feed grain markets and
market prospects. http://www.globalbioenergy.org/uploads/media/0806_Keith_
Collins_-_The_Role_of_Biofuels_and_Other_Factors.pdf (last visited 15/01/2012).
Du, X., Yu, C.L., Hayes, D.J., 2011. Speculation and volatility spillover in the crude oil and
agricultural commodity markets: a Bayesian analysis. Energy Econ. 33 (3), 497–503.
Enders, W., 2004. Applied Econometric Time Series, 2nd ed. John Wiley & Sons Inc.,
USA.
Gilbert, C.L., 2010. How to understand high food prices. J. Agric. Econ. 61, 398–425.
Hafner, C.M., Herwartz, H., 2006. A Lagrange multiplier test for causality in variance.
Econ. Lett. 93 (1), 137–141.
Hamilton, J.D., 2009. Understanding crude oil prices. Energy J. 30, 179–206.
Harri, A., Hudson, D., 2009. Mean and variance dynamics between agricultural com-
modity prices and crude oil prices. The Economics of Alternative Energy Sources
and Globalization: The Road Ahead Meeting, Orlando, FL, November 15–17, 2009.
Harri, A., Nalley, L., Hudson, D., 2009. The relationship between oil, exchange rates, and
commodity prices. J. Agric. Appl. Econ. 41 (2), 501–510.
Hassouneh, I., Serra, T., Goodwin, B.K., M. Gil, J.M., 2012. Non-parametric and parametric
modeling of biodiesel, sunflower oil, and crude oil price relationships. Energy Econ.
34 (5), 1507–1513.
665
Hong, Y., 2001. A test for volatility spillover with application to exchange rates. J. Econ.
103, 183–224.
Koop, G., Pesaran, M.H., Potter, S.M., 1996. Impulse response analysis in nonlinear mul-
tivariate models. J. Econ. 74, 119–147.
Krichene, N., 2008. Crude Oil Prices: Trends and Forecast. Working Paper 08/133. Inter-
national Monetary Fund, Washington.
Kristoufek, L., Janda, K., Zilberman, D., 2012. Correlations between biofuels and related
commodities before and during the good crisis: a taxonomy perspective. University
of California Center for Energy and Environmental Economics Working Paper
(http://www.uce3.ucsb.edu/WP_030.pdf last visited 20/06/2012).
Lazear, E., 2008. Responding to the global food crisis. Council of Economic Advisers Testi-
mony to the Senate Foreign Relations Committee. (May14,2008, http://georgewbush-
whitehouse.archives.gov/cea/lazear20080514.htmlS last visited 17/04/2009).
Mitchell, D., 2008. A Note on Rising Food Prices. Policy Research Working Paper
No. 4682. The World Bank, Washington, DC.
Nazlioglu, S., 2011. World oil and agricultural commodity prices: evidence from nonlinear
causality. Energy Policy 39 (5), 2935–2943.
Nazlioglu, S., Soytas, U., 2011. World oil prices and agricultural commodity prices:
evidence from an emerging market. Energy Econ. 33 (3), 488–496.
Nazlioglu, S., Soytas, U., 2012. Oil price, agricultural commodity prices and the dollar: a
panel cointegration and causality analysis. Energy Econ. 34, 1098–1104.
Pesaran, M.H., Shin, Y., 1998. Generalized impulse response analysis in linear multivar-
iate models. Econ. Lett. 58, 17–29.
Pindyck, R.S., Rotemberg, J.J., 1990. The excess co-movement of commodity prices.
Econ. J. 100, 1173–1189.
Rosegrant, M.W., Zhu, T., Msangi, S., Sulser, T., 2008. Global scenarios for biofuels:
impacts and implications. Rev. Agric. Econ. 30, 495–505.
Sari, R., Hammoudeh, S., Chang, C.L., McAleer, M., 2011. Causality between market
liquidity and depth for energy and grains. http://ssrn.com/abstract=1817022
(last visited 02/2012.).
Schmidhuber, J., 2007. Biofuels: An Emerging Threat to Europe's Food Security?
(Notre Europe. http://www.notre-europe.eu/uploads/tx_publication/Policypaper-
Schmidhuber-EN.pdf last visited 07/2011).
Serra, T., 2011. Volatility spillovers between food and energy markets: a semiparametric
approach. Energy Econ. 33 (6), 1155–1164.
Serra, T., Zilberman, D., Gil, J.M., Goodwin, B.K., 2011. Nonlinearities in the U.S. corn–
ethanol–oil–gasoline price system. Agric. Econ. 42, 35–45.
Tyner, W.E., 2008. The U.S. ethanol and biofuels boom: its origins, current status, and
future prospectus. BioScience 58, 646–653.
Tyner, W.E., Taheripour, F., 2008. Policy options for integrated energy and agricultural
markets. Rev. Agric. Econ. 30 (3), 387–396.
Yang, J., Qiu, H., Huang, J., Rozelle, S., 2008. Fighting global food price rises in the devel-
oping world: the response of China and its affect on domestic and world markets.
Agric. Econ. 39, 453–464 (supplement).
Zhang, Z., Lohr, L., Escalante, C., Wetzstein, M., 2010. Food versus fuel: what do prices
tell us? Energy Policy 38, 445–451.