Professional Documents
Culture Documents
The effect of global oil price shocks on China’s agricultural commodities
PII: S0140-9883(15)00221-2
DOI: doi: 10.1016/j.eneco.2015.07.012
Reference: ENEECO 3127
Please cite this article as: Zhang, Chuanguo, Qu, Xuqin, The effect of global oil
price shocks on China’s agricultural commodities, Energy Economics (2015), doi:
10.1016/j.eneco.2015.07.012
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
Title Page
P T
commodities
RI
SC
Chuanguo Zhang, Xuqin Qu
NU
MA
Corresponding author: Chuanguo Zhang
Email: cgzhang@xmu.edu.cn.
PT
CE
Co-author: Xuqin Qu
AC
Phone: 86-592-18106989932
E-mail: 798307772@qq.com
1
ACCEPTED MANUSCRIPT
Abstract: This paper studied the effect of global oil price shocks on agricultural
commodities in China, including strong wheat, corn, soybean, bean pulp, cotton and
P T
process and jump process. We not only separated oil price shocks into positive and
RI
negative categories to identify different effects on agricultural commodities in
SC
continuous process, but also investigated how jump behavior influenced these
agricultural commodities. We found that the oil price was characterized by volatility
NU
clustering and jump behavior. At the same time, oil price shocks had different
MA
effects on agricultural commodities. In addition, the shocks on most agricultural
commodities were asymmetric. Only natural rubber was under the influence of the
ED
jump intensity of the oil price, in contrast to strong wheat, corn, soybean, bean pulp
PT
and cotton.
2
ACCEPTED MANUSCRIPT
1. Introduction
As the most important raw material and basic energy in the world, crude oil is of
T
vital importance to socio-economic development and stability. China has been listed
P
the second-largest oil-consumption country in the world, second only to the U.S. since
RI
2003 and afterwards has become the world's largest net oil-importing country since
SC
September 2013. External dependency on oil has improved in line with the growth of
NU
economy, mirroring the growing interdependence between China and the international
oil market. China’s oil consumption was 272.74 million tons in 2003 and reached
MA
507.40 million tons in 2013–an increase of 86.04%. Meanwhile, China's crude oil
ED
imports rose from 91.02 million tons in 2003 to 281.92 million tons in 2013,
33.37% in 2003. This figure rapidly climbed to 55.58% in 2013. Due to high external
CE
dependency on oil, global oil price fluctuations will affect relevant industries through
a variety of price transmission mechanisms and further bring risks and challenges for
AC
Trading is brisk on the global oil market; the oil price is not only affected by the
supply–demand relationship but also by other external market information. The global
oil price has displayed violent fluctuations under the impact of Iraq war and the
spreading subprime crisis during the 21st century. For example, the Brent crude oil
spot price was $46.53 per barrel on September 23, 2004. Then from September 2004
to July 2008, the global oil price has been rising. On July 3, 2008, it even rose as high
as $143.95 per barrel–up around 209.37%. However, shortly after the peak, due to
3
ACCEPTED MANUSCRIPT
disturbance of the exceptional information, the Brent crude oil spot price dropped
sharply to $33.73 per barrel on December 26, 2008–down about 76.57%. Henceforth,
it began to enter into a new motion cycle with relatively moderate volatilities. The
P T
Brent crude oil spot price mainly hovered at around $100 per barrel from February
RI
2011 to August 2014. Yet, soon after, there has appeared another drastic plunge in the
SC
global oil price since October 2014.
NU
society (He et al., 2012). Oil price shocks will not only affect the macroeconomy, but
MA
also will bring risks for industrial development. Some people blame the rise in the
oil price for the large increase in inflation. And as production input factors for most
ED
industries, the skyrocketing oil price will inevitably drive up the costs of production
PT
patterns lead to the changes in global oil prices having greater impacts on agriculture
AC
fertilizers and so on. According to the statistics from China’s Statistical Bureau, from
2002 to 2013, the price indices of chemical fertilizers and oils for farm machinery
rose more than 90%. In addition, the corn-based biofuel production also makes the
corn price be more vulnerable to fluctuations in the oil price. According to the study
of Ciaian and Kancs (2011), due to the expansion of biofuels from 2004 to 2008, the
interdependencies between the fuel and agricultural prices was expected to be stronger.
Vacha et al. (2013) also found that the connection between ethanol and corn was
4
ACCEPTED MANUSCRIPT
at the mercy of oil price fluctuations, the whole national economy will be heading for
P T
serious trouble. Thus it is necessary to investigate how global oil price shocks affect
RI
China’s agricultural commodities.
SC
Currently, there exists a great deal of research investigating the spillover effect of
the crude oil price. Some scholars are concerned about how oil price shocks affect the
NU
macroeconomy. In their opinion, there existed a negative correlation between the oil
MA
price and macroeconomy (Baláž and Londarev, 2006; Hamilton, 2005). Meanwhile,
others launch a more detailed analysis about the effect of oil price fluctuations on
ED
(Jimenez-Rodriguez, 2008) and the stock market (Cunado and Perez De Gracia, 2014)
and so on. However, most research is mainly conducted from the perspective of
CE
developed countries. Only a few scholars focus on China at the industry level. For
AC
example, Wang and Zhang (2014) researched the influence of oil price fluctuations on
China’s fundamental industries. Up to now, there are few studies involving the impact
This paper mainly discusses how oil price shocks impact China’s agricultural
agricultural commodities: strong wheat, corn, soybean, bean pulp, cotton and natural
rubber. We apply ARJI-GARCH model to investigate how oil price shocks, especially
the jump behavior of the oil price, affect China’s agricultural commodities. We divide
5
ACCEPTED MANUSCRIPT
oil price shocks into positive and negative categories to investigate the asymmetric
This work is different from previous research in three aspects: Firstly, the paper
P T
analyzes China’s six kinds of specific agricultural commodities: strong wheat, corn,
RI
soybean, bean pulp, cotton and natural rubber, unlike previous studies focusing on
SC
macroeconomy or the whole fundamental industries. Secondly, we apply
ARJI-GARCH model, rather than traditional VAR model or the impulse response
NU
function to investigate how oil price shocks, especially the jump behavior of the oil
MA
price, affect China’s agricultural commodities. Finally, we investigate the asymmetric
2. Literature review
PT
Accompanied by severe oil price fluctuations in recent years, the influence of oil price
AC
shocks is being paid more attention and has gradually become a popular topic for
study. The existing research mainly focuses on the effect of oil price shocks on
different development levels. In the U.S., nine out of ten recessions may have arisen
from the rise in oil prices since World War II (Hamilton, 2005). In Turkey, high oil
6
ACCEPTED MANUSCRIPT
prices had very harmful effects on economy. It could not only lead to the decline in
output and consumption but also arouse the deterioration of the net foreign asset
position (Aydın and Acar, 2011). In Thailand, Rafiq et al. (2009) pointed out that
P T
there existed unidirectional causality between oil price volatility and some
RI
macro-economic variables from 1993 to 2006 in terms of the Granger causality test.
SC
In China, according to Ou et al. (2012), China’s macroeconomy was less independent
on the WTI price shocks apart from China’s foreign trade and stock market in the
NU
light of the price transmission mechanism. In Nigeria, most macroeconomic variables
MA
were under weak influence of oil price shocks (Iwayemi and Fowowe, 2011). In
general, the effects of oil price shocks on the macroeconomy mainly involve three
ED
aspects. First, the potential effects of oil price fluctuations are on investment and
PT
output. An increase in the oil price would negatively affect the investment and output
in China (Tang et al., 2010). Second, oil price shocks can result in changes in
CE
unemployment rates. From 1947 to 1995, Uri (1996) suggested that there existed
AC
the United States. At last, oil price shocks keep in close touch with inflation.
Cavalcanti and Jalles (2013) found that oil price shocks accounted for a larger fraction
As the connection between the international crude oil market and relevant
industries is becoming closer, some scholars have begun to focus on this topic from
the aspect of industry, and point out that oil price volatilities obviously have different
7
ACCEPTED MANUSCRIPT
industries responded to oil price shocks depended on certain industries and the
Due to owning similar properties with crude oil, the relationship between energy
P T
commodities and crude oil has been paid more attention. Scholtens and Yurtsever
RI
(2012) found that mining, oil and gas industries were positively affected by oil price
SC
increases while negatively affected by falling oil prices. According to the study of
Ewing et al. (2002), the oil sector was impacted by the natural gas sector by direct and
NU
indirect volatility transfer effect, while the natural gas sector was directly affected by
MA
events in its own sector and indirectly under influence of the oil sector. Panagiotidis
and Rutledge (2007) suggested that there existed a cointegrating relationship between
ED
the gas and oil price in the UK during the sample period (1996-2003) in terms of a
PT
(2005) found that the relationship between oil prices and equity values in the non-oil
CE
and gas sectors was weak. Mohammadi (2009) also found that the long-term
AC
connections between electricity and crude oil and/or natural gas prices were
insignificant.
industries. One is the metals industry. Hammoudeh and Yuan (2008) found that oil
price volatilities negatively impacted gold and silver volatilities except copper
volatility. According to Narayan et al. (2010), the price in the gold market was at the
mercy of the price in the crude oil market and vice versa. Wang and Chueh (2013)
also showed that in the short term both gold and crude oil prices positively influenced
8
ACCEPTED MANUSCRIPT
each other. However, Soytas et al. (2009) found that global oil price had no impact on
precious metal prices in Turkey in the long run and vice versa, although there existed
transitory positive initial impacts of innovations in oil prices on gold and silver
P T
markets in the short run.
RI
The other is the agricultural industry. Some studies pointed out that there existed
SC
information transmission from the global oil price to several kinds of agricultural
NU
2009; Nazlioglu and Soytas, 2012). Oil price shocks have different effects on
MA
agricultural commodity prices in different periods. Before the food crisis in
2006–2008, oil price shocks only accounted for a minor friction of variations in
ED
agricultural commodity prices. However after the crisis, oil price shocks contributed
PT
(Wang et al., 2014). As for the mechanism by which oil price shocks influence the
CE
agricultural commodity prices, there are also two opinions. On the one hand, the oil
AC
price changes affected agricultural commodity prices by means of increasing the costs
transportation costs. On the other hand, the ongoing growth of corn-based ethanol
production induced a higher derived demand for corn or soybeans which finally
resulted in higher prices of corn and soybeans (Chen et al., 2010). Based on nonlinear
causality analysis, Nazlioglu (2011) found that there was a persistent unidirectional
nonlinear causality from the price in the oil market to the price in the corn and
soybean markets. Mutuc et al. (2010) concluded that the global oil demand shocks
9
ACCEPTED MANUSCRIPT
originating from the improved economic activity played an important role in the
cotton market while oil supply shocks and precautionary oil demand shocks did not
contribute much to cotton price variation. However, there also existed different
P T
opinions on the question. For example, according to linear causality analysis,
RI
Nazlioglu and Soytas (2011) found that the oil price and agricultural commodity
SC
prices did not influence each other in Turkey, which supported the evidence on the
neutrality hypothesis.
NU
In terms of analytical methodologies, a number of literature adopted the
MA
Bayesian analysis (Du et al., 2011), the panel cointegration and causality analysis
(Farzanegan and Markwardt, 2009; Nazlioglu and Soytas, 2012), the detrended
ED
Beckmann and Czudaj (2014) applied a GARCH-in-mean VAR model to study the
short term effects of oil volatility spillovers on the agricultural futures market. Haixia
and Shiping (2013) employed the univariate EGARCH model and the
China’s crude oil, fuel ethanol and corn markets. Cunado and Perez De Gracia (2014)
used VAR (Vector Autoregressive) and VECM (Vector Error Correction Model) to
investigate how oil price shocks impacted the stock returns in 12 oil importing
10
ACCEPTED MANUSCRIPT
European economies. In order to better describe the continuous volatility process and
the jump behavior of the oil price, Chan and Maheu (2002) adopted the jump model
along with a GARCH model to analyze the time series characteristics of daily stock
P T
returns during a 72 year period. In addition, Maheu and McCurdy (2004) presented a
RI
mixed GARCH–jump model to study the individual security returns. Lee et al. (2010)
SC
employed a Component-ARJI model with structural break analysis to investigate the
conditional variance and jump intensity of the oil price. To investigate the asymmetric
NU
effect of oil price shocks on stock returns and explore the importance of structure
MA
changes in this dependency relationship, Chiou and Lee (2009) implemented the ARJI
such as the EU and the United States, while only a few efforts have been made
concerning China. Meanwhile, most studies pay more attention to the macroeconomy
CE
and industry rather than specific commodities. For example, Ou et al. (2012)
AC
investigated how China’s macroeconomy responded to global oil price shocks. Zhang
and Chen (2014) focused on how global oil price shocks impacted China’s bulk
commodity markets and fundamental industries. With high external dependency on oil,
the relationship between China and the global oil market is becoming closer than
before. The global oil price fluctuations will exert an influence first of all on some
11
ACCEPTED MANUSCRIPT
11, 2001, the agricultural commodity prices are not only affected by traditional factors
such as cost factors, but also impacted by external factors, especially the global oil
P T
agricultural commodity prices for a long time. Therefore, it is necessary to analyze
RI
how global oil price shocks affect China’s agricultural commodities.
SC
3. Methodology
the process of analyzing the time series data (Pourahmadi, 1992). In general, we
ED
would employ the ARMA(p, q) model to estimate parameters and select a suitable
PT
dimension of the moving average component of the model (Ives et al., 2010). The
ARMA model of order (p, q) can be expressed as follows (Box and Jenkins, 1976):
AC
A number of literature has shown that the GARCH model is appropriate for
analyzing the conditional variance, therefore we combine the jump model with a
GARCH model (Chan and Maheu, 2002). Define the history of returns as the
12
ACCEPTED MANUSCRIPT
ARJI-GARCH model combining the continuous process with the jump behavior of
P T
RI
SC
Eq (2) is the conditional mean function of the Brent crude oil returns. denotes
1986). denotes the white noise sequence. and denote the independent
ED
Let denote the discrete counting process governing the numbers of jumps.
Suppose it follows the Poisson distribution and is conditional on the information set
CE
with the parameter > 0, the probability density function is as follows (Chan and
AC
Maheu, 2002):
Consider the ARJI (r, s) model, both parameters are equal to 1 in our empirical
analysis. Let denote the conditional jump intensity which is equal to the
conditional expected value of the counting progress (Chan and Maheu, 2002):
Having observed and according to the Bayes’s rule, we can calculate the
T
ex-post probability of the occurrence of j jumps at time t as follows:
P
RI
The conditional density of returns is as follows (Chan and Maheu, 2002; Maheu
SC
and McCurdy, 2004):
NU
MA
To derive our model’s conditional variance, let denote the jump
into two separate parts: the diffusion-induced component and the jump-induced
Given sample size T, the log-likelihood function can be written as follows (Tang
et al., 2010):
P T
Let denote all the parameters to be estimated.
RI
SC
3.3 The ARMA-GARCH model
In order to examine how positive and negative categories of the Brent crude oil
NU
returns affect agricultural commodities, we decompose the Brent crude oil returns into
MA
positive and negative categories. What’s more, to observe the response of agricultural
commodities to the jump behavior of Brent crude oil returns, we also integrate the
ED
( , 0) denotes the positive crude oil returns, and denotes the jump intensity series
The error term usually follows a GARCH progress of order (1, 1) in the
series.
15
ACCEPTED MANUSCRIPT
T
wheat, corn, soybean, bean pulp, cotton and natural rubber. It is well known that data
P
frequency plays fundamental roles in the process of price transmission from oil prices
RI
to agricultural prices (Nazlioglu, 2011). Therefore, our study adopts 2257 daily
SC
observations from September 23, 2004 to April 3, 2014. The six kinds of agricultural
NU
commodity price indices are derived from the official database of China's Webstock
(http://www.wenhua.com.cn/) and the Brent crude oil spot price is derived from the
MA
U.S. Energy Information Administration (EIA). To ensure the consistency and
ED
comparability among different data, we select all data appearing at the same date in
the seven time series. Returns are defined as follows and parameters are estimated by
PT
As shown in Table 1, the standard deviation of the Brent crude oil index is the
largest one, indicating that it has the most violent fluctuations. In addition to the
strong wheat index owning the positive skewness, the other commodity indices are
negatively skewed. As for the kurtosis, higher values of the kurtosis indicate that the
distributions of all time series have characteristics of sharp peaks and fat tails.
According to the Jarque-Bera test, all time series reject the normal distribution
assumption at the 1% level. In view of the property of the skewness and kurtosis, we
16
ACCEPTED MANUSCRIPT
can confirm that none of the time series follow the normal distribution. And in light of
the Ljung-Box Q and Q2 statistics, all series own the property of conditional
P T
model.
RI
Table 1
SC
Summary statistics (2004-2013)
Strong Natural
Variable Brent oil Corn Soybean Bean pulp Cotton
Wheat Rubber
NU
Mean 0.0360 -0.0283 0.0000 0.0122 0.0359 -0.0104 -0.0082
Std. 2.2014 0.7227 0.6942 1.1801 1.3799 1.0374 1.9035
Skewness -0.1635 0.2508 -0.0381 -0.2796 -0.1681 -0.1740 -0.1484
MA
Kurtosis 10.3887 8.6314 9.6587 6.8821 5.0683 9.6912 3.7952
*** *** *** *** *** ***
Jarque-Bera 5141.8270 3004.6080 4168.3540 1446.0510 412.7313 4219.9400 67.7135***
Q(10) 27.2695*** 20.0143** 53.7559*** 30.6272*** 24.4874*** 23.9341*** 33.8686***
Q2 (10) 342.0915*** 158.3536*** 232.0698*** 1164.5370*** 703.8324*** 1592.7290*** 433.2582***
ED
Note: ***, ** indicate statistical significance at the 1 and 5 percent level of significance, respectively.
PT
Since most time series variables have a unit root, without considering their unit
root property, we cannot directly use these data (Fang and You, 2014). To examine the
CE
stationary of these data, we conduct the unit root test by the traditional Augmented
AC
Dickey–Fuller (ADF) test (Agiakloglou and Newbold, 1996), the Phillips and Perron
(Kwiatkowski et al., 1992). The null hypothesis of the ADF and PP tests with the
constant term and linear trend is that time series is not stable. In other words, there
exists unit root. KPSS test is carried out after removing the trend term of the original
time series, the null hypothesis is that time series is stable in the long term. If the ADF
and PP tests cannot reject the null hypothesis and KPSS test rejects the null
17
ACCEPTED MANUSCRIPT
As shown in Table 2, contrary to the KPSS test, the seven price indices series
cannot reject the null hypothesis according to the ADF and PP tests, suggesting that
all of the price indices series are not stable. However, all the returns indices series
P T
reject the null hypothesis in terms of the ADF and PP tests, in contrast to the results of
RI
the KPSS test. Therefore, we can infer that all of the returns indices series are stable
SC
and do not exist unit root.
NU
Table 2
Unit root and stationary tests
** *** ***
Cotton -1.6452 -1.8590 0.7145 -46.3791 -47.0110 0.1047
*** *** ***
Natural Rubber -1.5240 -1.6630 1.1957 -49.5440 -49.5254 0.0950
Note: (1) ADF and PP statistics adopt t-Statistic for judgment, while KPSS statistics adopt LM-Statistic for
CE
judgment. (2) ***, ** indicate statistical significance at the 1 and 5 percent level of significance, respectively.
AC
As shown in Fig. 1 and Fig. 2, from 2004 to 2014, the fluctuations of the Brent
crude oil spot price are severe. And when oil price jumps appear, the volatility rate of
the oil price will change dramatically. According to Fig. 3 we can find that the
volatility clustering phenomenon of the six indices series are obvious, a large (small)
volatilities in the six indices series can be forecast to some extent (Haixia and Shiping,
2013). Thus it indicates that the GARCH effects exist in the six indices series.
18
ACCEPTED MANUSCRIPT
160 20
Brent Crude Oil Spot Price ($) Brent Crude Oil Growth Rate (%)
140 15
120 10
100 5
80 0
T
60 -5
P
40 -10
20 -15
RI
0 -20
2008
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2004
2005
2006
2007
2009
2010
2011
2012
2013
SC
Fig. 1 Time Series plots of Brent oil.
25
Volatility of Brent Crude Oil (%)
NU
MA
20
15
10
ED
5
PT
0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
the diffusion-induced component the jump-induced component the total volatility of crude oil
CE
19
ACCEPTED MANUSCRIPT
1200 8
Strong Wheats Index (RMB) Strong Wheats Growth Rate (%)
1000 6
800 4
600 2
T
400 0
P
200 -2
RI
0 -4
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
SC
1400 6
Corns Index (RMB) Corns Index Growth Rate (%)
NU
1200 4
1000
2
800
0
MA
600
-2
400
200 -4
0 -6
ED
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2007
2004
2005
2006
2008
2009
2010
2011
2012
2013
PT
1600 8
Soybeans Index (RMB) Soybeans Index Growth Rate (%)
1400 6
1200 4
CE
1000 2
800 0
600 -2
AC
400 -4
200 -6
0 -8
2012
2004
2005
2006
2007
2008
2009
2010
2011
2013
2011
2012
2004
2005
2006
2007
2008
2009
2010
2013
2500 8
Bean pulps Index (RMB) Bean pulps Index Growth Rate (%)
6
2000
4
1500 2
0
1000 -2
-4
500
-6
0 -8
2004
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2005
2006
2007
2008
2009
2010
2011
2012
2013
20
ACCEPTED MANUSCRIPT
1400 8
Cottons Index (RMB) Cottons Index Growth Rate (%)
1200 6
1000 4
2
800
0
600
T
-2
400 -4
P
200 -6
RI
0 -8
2007
2010
2013
2004
2005
2006
2008
2009
2011
2012
2010
2004
2005
2006
2007
2008
2009
2011
2012
2013
SC
2500 8
Natural Rubbers Index (RMB) Nature Rubbers Growth Rate (%)
NU
6
2000
4
1500 2
0
MA
1000
-2
-4
500
-6
0 -8
ED
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2012
2004
2005
2006
2007
2008
2009
2010
2011
2013
PT
4.2 Application of the ARMA model to the Brent oil returns (2004-2014)
AC
According to the results shown in Table 3, in addition to the constant term, all
we find that the ARMA (2, 2) model has the minimum Akaike information value, so
Table 3
ARMA model on Brent oil returns (2004–2014).
Variable Coefficient Std. Error t-Statistic Prob.
0.0357 0.0466 0.7667 0.4434
***
-0.6908 0.0167 -41.3142 0.0000
-0.9651*** 0.0161 -59.8933 0.0000
***
0.7109 0.0167 42.4737 0.0000
21
ACCEPTED MANUSCRIPT
T
4.3 Application of the ARJI-GARCH model to the Brent crude oil returns
P
(2004-2014)
RI
According to the results in Table 4, the coefficients of ARMA (2, 2) model ( ,
SC
, , ) are significant at the 1% level except that the constant term is
NU
significant at the 5% level. The coefficients of GARCH (1, 1) model ( , , ) are
significant at the 10%, 5%, 1% level respectively, indicating that the Brent crude oil
MA
has significant GARCH effect. According to Fig. 2, it is not hard to find that the
ED
volatility rate of the Brent crude oil index returns maintained at a higher level in 2008.
However, during the period from 2004 to 2007 and the period from 2009 to 2014, the
PT
volatility rate is relatively lower, indicating that the time series of the Brent crude oil
CE
employ the GARCH model. As shown in table 3, the mean of the jump size ( ) is
AC
significant at the 10% level and the variance ( ) is significant at the 1% level,
indicating once the international oil market is under influence of some abnormal
information, the oil price will respond immediately, or even will appear some extreme
movements, such as price jumps. In addition, the coefficient of the jump intensity ( )
is significant at the 10% level and the other two coefficients ( , ) are significant at
the 1% level, demonstrating it is reasonable to employ the ARJI model. The positive
suggesting that the current jump intensity ( ) is mainly affected by the previous jump
22
ACCEPTED MANUSCRIPT
intensity. According to Fig. 4, we can find that the jump intensity of the Brent crude
oil index returns mainly stays between 0 and 1. However, once the international oil
P T
change dramatically. For example, under impact of the spreading subprime crisis in
RI
2008, the oil price fluctuated violently and the jump intensity reached 1.4860 on
SC
January 12, 2009. So it is clear that the jump intensity of oil price is time-varying and
NU
statistics, the standardized residual series cannot reject the null hypothesis,
MA
indicating that the time series do not exist correlation, thus it is reasonable to employ
Table 4
ARJI–GARCH on Brent oil returns (2004–2014).
PT
***
1.2230 0.1890 6.4692 0.0000
***
-0.6617 0.1789 -3.6979 0.0002
***
-1.2263 0.1913 -6.4092 0.0000
AC
***
0.6675 0.1808 3.6912 0.0002
*
0.0054 0.0032 1.7204 0.0854
**
0.0103 0.0049 2.1129 0.0346
***
0.9814 0.0063 155.3198 0.0000
2.3583*** 0.3722 6.3358 0.0000
*
-0.3548 0.2054 -1.7275 0.0841
*
0.0049 0.2054 1.7236 0.0848
***
0.9820 0.0093 106.0322 0.0000
***
0.1915 0.0719 2.6654 0.0077
2.3619 [0.7971]
Note: (1) , , , , are the parameters of ARMA(2,2), defined in Eq.(2). (2) , and are the
parameters of GARCH (1,1), defined in Eq.(3). (3) and are the mean and variance of the jump size
respectively. (4) , , are the parameters of jump intensity , defined in Eq.(6). (5) is the Ljung-Box
test statistics for serial correlation in the squared standardized residuals with 5 lags and the value in the square
bracket indicates the significance level. (6) ***, **, * indicate statistical significance at the 1, 5 and 10 percent level
of significance, respectively.
23
ACCEPTED MANUSCRIPT
1.6
Jump Intensity of Crude Oil
1.4
1.2
1.0
T
0.8
P
0.6
RI
0.4
0.2
SC
0.0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
NU
4.4 Application of the ARMA-GARCH model to agricultural commodity indices
MA
returns
six indices returns with 20 lags are not significant at the 1% level, indicating that the
PT
six residual series do not exist with autocorrelation, so it is appropriate and effective
CE
significant at the 5% level, indicating that the six kinds of agricultural commodities
significant at the 10% level at least, demonstrating that all agricultural commodities
are affected by the positive and negative oil price shocks. In order to compare the
coefficients and , the likelihood ratio (LR) test is adopted. The null hypothesis
model, where denotes all the parameters to be estimated. The LR statistics follow
cannot reject the null hypothesis, that is to say, , otherwise, we will reject the
P T
null hypothesis. According to the results from the LR test in Table 5, except for the
RI
cotton index returns, the other five agricultural commodity indices returns reject the
SC
null hypothesis at the 10% level at least, indicating that oil price shocks have
asymmetric effects on strong wheat, corn, soybean, bean pulp and natural rubber. By
NU
comparison, we find that only the natural rubber index returns are more sensitive to
MA
positive oil price shocks, in contrast with strong wheat, corn, soybean and bean pulp
indices returns, demonstrating that most agricultural commodities are more sensitive
ED
In addition, we find that only natural rubber index returns are affected by the
current and previous jump intensity at the 1% level, while the rest of the agricultural
CE
commodities indices returns fail to respond to the jump intensity of the crude oil price.
AC
The coefficients of and are -2.5571 and 2.4323, respectively, indicating that
natural rubber is negatively affected by the current jump intensity while positively
affected by the previous jump intensity. Thus we can infer that after overreacting to
the current jump intensity of the oil price, natural rubber tends to adjust back to a
Table 5
AR-GARCH model on the agricultural commodity markets (2004-2014).
parameter strong wheat corn soybean Bean pulp cotton Natural rubber
-0.0040 0.0102 0.0017 0.0365 -0.0116 -0.0093
** **
0.1146 0.2094 0.0972 0.2123 0.4577 0.4651***
25
ACCEPTED MANUSCRIPT
T
0.0271*** 0.0263** 0.0928*** 0.1147*** 0.0451 *** 0.1704 ***
0.0327*** 0.0405*** 0.1056*** 0.1520*** 0.0251* 0.1284***
P
0.0017 ** 0.0063*** 0.0125*** 0.0160*** 0.0114*** 0.0609***
RI
0.0543*** 0.0801*** 0.0815*** 0.0506*** 0.1070*** 0.0654***
*** *** *** *** ***
0.9444 0.9106 0.9113 0.9402 0.8838 0.9179***
SC
11.6420 14.3950 11.8701 15.5979 21.3123 21.6821
[0.9278] [0.8099] [0.9205] [0.7412] [0.3790] [0.3580]
-2154.5004 -2095.9088 -3168.5654 -3636.8160 -2695.8941 -4462.6025
NU
-2158.9864 -2100.4053 -3170.1386 -3639.6391 -2696.8250 -4464.1992
*** *** * **
8.9720 8.9930 3.1464 5.6462 1.8618 3.1934*
Note: (1) , , , , are parameters of ARMA(2, 2), defined in Eq.(14). (2) and are
MA
coefficients of the current and the first-order lag jump intensity of the oil price respectively, defined in Eq.(14). (3)
and are coefficients of the positive and negative oil price shocks respectively, defined in Eq.(14). (4) ,
and are parameters of GARCH (1, 1), defined in Eq.(16). (5) is the Ljung-Box test statistics for
serial correlation in the squared standardized residuals with 20 lags and the value in the square bracket indicates
ED
the significance level. (6) is the maximum likelihood value in the unconstrained model, is the
maximum likelihood value in the constrained model ( ), LR denotes the likelihood ratio, where LR = 2
***, **, *
( - ). (7) indicate statistical significance at the 1, 5 and 10 percent level of significance,
PT
respectively.
CE
5. Discussions
AC
5.1 The oil price is characterized by volatility clustering and jump behavior
The volatility clustering means that a large (small) volatility is always followed
by large (small) volatilities (Haixia and Shiping, 2013). The jump behavior suggests
that the oil price perhaps will sharply rise up or fall down when affected by
unexpected events. Askari and Krichene (2008) also found that oil price dynamics had
characteristics of high volatility, high intensity jumps and strong upward drift during
2002-2006. Yuan et al. (2014) pointed out that the oil price series showed
26
ACCEPTED MANUSCRIPT
(2014), the volatility clustering feature resulted from investors’ different response to
oil price changes. In our opinions, the volatility clustering phenomenon of global oil
price is closely related with the development of oil futures in recent years. Oil futures
P T
brings oil and finance together, and the financial activities gradually integrate into oil
RI
trading activities. Once the oil price appears small fluctuations, in order to avoid risks,
SC
hedging transactions will be implemented more frequently in the oil futures market.
As a result, such behaviors will lead to further volatilities in the oil price and finally
NU
perhaps trigger volatility clustering phenomenon. As for the jump behavior, in the past
MA
few years, the international environment has become more complex than before,
which brings more uncertainty for the global oil price. In particular, unexpected
ED
events may lead to the jump behavior appearing in the process of oil price volatility.
PT
For example, the oil price took a nosedive in the crisis between 2008 and 2009 and
accordingly the jump intensity of the oil price was as high as 1.4860 on January 12,
CE
2009.
AC
5.2 The oil price shocks have different effects on agricultural commodities
In the paper, the six kinds of agricultural commodities can be divided into two
categories: One is the food crops, including strong wheat and corn; the other is the
cash crops, including soybean, bean pulp, cotton and natural rubber. In
comparison with the two types of agricultural commodities, we find that the impact of
oil price shocks on cash crops is stronger than that on food crops. The discrepancy
perhaps can be explained as follows: (1) The external dependencies on imports are
different. The food crops are less dependent on foreign imports than the cash
27
ACCEPTED MANUSCRIPT
crops. According to relevant data from the Information Center of National Food and
Oil Organization, in 2011 the self-sufficiency rate of wheat and corn were 98.9% and
99.1%, respectively. Both staple food production capacities can basically meet the
P T
domestic consumption needs, therefore the food crops are less affected by external
RI
factors, while the external dependency of cash crops is relatively higher. According to
SC
China's National Bureau of Statistics, in 2013, China imported 63.38 million tons of
soybeans, accounting for 80% of China’s soybean market. At the same time, based on
NU
data from China Feed Industry Association, the production of bean pulp in China
MA
mainly relies on imported soybeans and its external dependency was as high as 82%
cotton is unstable in China and its external dependency reached 30% in 2010.
PT
natural rubber in 2012 was closed to 80% in China. Thus the high external
CE
dependencies lead to the cash crops being more susceptible to the influence of
AC
external factors, such as oil price shocks. (2) The cost-push effects of the oil price are
different. For food crops, the impact of oil price shocks is mainly reflected in the cost
in China is lower than that in developed countries. Food crops are not sensitive to
changes in the cost of inputs, therefore, oil price shocks have fewer effects on food
crops. For cash crops, oil price shocks are passed to them mainly by the costs from
transportation as well as the related production and processing process. Thus cash
crops are more closely associated with oil price shocks. (3) The linkages with
28
ACCEPTED MANUSCRIPT
economic activity are different. As the fundamental means of subsistence, food crops
are closely related to people's daily lives. While as industrial raw material, cash crops
have a higher economic value and commodity rate, closely related to economic
P T
activity. Compared with people's daily lives, economic activity is more susceptible to
RI
oil price fluctuations.
SC
5.3 The oil price shocks on most agricultural commodities are asymmetric
NU
According to the results from the LR test, we find that except of cotton, the rise
and fall in the oil price have different effects on the other five agricultural
MA
commodities. Therefore, we can infer that oil price shocks have asymmetric effects on
ED
most agricultural commodities. For most agricultural commodities, they are more
sensitive to the fall in the oil price, such as strong wheat, corn, soybean and bean pulp,
PT
while natural rubber are more easily affected by the rise in the oil price. Generally,
CE
there are three views on the issue due to different research objectives and method: the
first view is that oil price shocks have asymmetric effects (Iwayemi and Fowowe,
AC
2011; Beckmann and Czudaj, 2014). The second one is that actually the asymmetric
effects are limited (Scholtens and Yurtsever, 2012; Kilian and Vigfusson, 2009). The
last one thinks that the neutrality hypothesis is valid (Nazlioglu and Soytas, 2011). As
for the reasons for such asymmetries, Iwayemi and Fowowe (2011) attributed it to the
reallocation effects and adjustment costs. According to Zhang and Chen (2011),
people were sensitive to oil price volatilities due to irrationality of many individual
investors in China which in turn incurred more speculations in the market. In our
study, the asymmetric effects of oil price shocks on most agricultural commodities
29
ACCEPTED MANUSCRIPT
perhaps can be attributed to the following explanations: on the one hand, the rise in oil
the price will lead to an increase in the price of agricultural commodities by raising
the cost of materials used for the products. At the moment, the irrational investors are
P T
inclined to trade in the stocks, then will be easily hung up in the high price. On the
RI
other hand, when the oil price falls, the price of agricultural commodities tends to
SC
decline, and some investors are anxious to trade out stocks to recoup their losses.
Such operations will inevitably increase the volatility in agricultural futures markets.
NU
So the irrational behavior of investors will lead to the oil price shocks showing
MA
asymmetric impacts.
5.4 The jump intensity of the oil price has different influence on agricultural
ED
commodities
PT
by the jump intensity of the oil price. This conclusion is supported by Wang and
Zhang (2014), who also pointed out that the jump intensity of the oil price had
AC
different impacts on the four commodity markets. We can see that the current jump
intensity ( ) of the oil price has a negative effect on the natural rubber index returns.
As the jump intensity of the oil price increases, the natural rubber index returns tend
to decline. While the first-order lag jump intensity ( ) of the oil price has a
positive impact on natural rubber index returns, indicating that natural rubber index
returns will react to the current jump intensity of the oil price positively and make an
opposite adjustment in the next period. Only natural rubber is affected by the jump
intensity of the oil price, there maybe exist two reasonable explanations: (1)
30
ACCEPTED MANUSCRIPT
Compared with the other agricultural commodities, natural rubber is widely used in
industry and has industrial property, and fluctuations in its futures price are broadly
consistent with the changes in economic activities. As the oil price jumps upwards,
P T
the economy will decline, resulting in the fall in the natural rubber price. Conversely,
RI
when the oil price jumps downwards, it is beneficial to the growth of the economy
SC
and the increase of the natural rubber price. (2) For the other five agricultural
NU
Exchange are mainly engaged in agricultural commodity futures trading, there are
MA
various kinds of agricultural commodities being traded on the market. The two
trading activities to improve futures market liquidity and promote the futures market,
PT
which finally contributes to hedging against oil price risk. In contrast, as a raw
material, the natural rubber is listed on Shanghai Futures Exchange with copper,
CE
aluminum and other industrial products. There is only one kind of agricultural
AC
when the jump intensity of the oil price increases, the crude oil market will become
more volatile. For natural rubber, once it is affected by the current jump intensity of
the oil price positively, it needs to make an opposite adjustment in the next period. In
a word, the jump behavior of the oil price undoubtedly will increase the risk of natural
rubber trade. Therefore, relevant departments should further enrich the kinds of
31
ACCEPTED MANUSCRIPT
agricultural commodity futures and make the agricultural commodity futures market
more active.
T
6. Conclusion
P
RI
In the paper, we investigate the impact of global oil price shocks on six kinds of
SC
agricultural commodities in China by adopting the ARJI-GARCH model and
ARMA-GARCH model. The global oil price fluctuations are divided into two parts:
NU
one is the smooth fluctuation under the influence of daily information, and the other is
MA
the jump behavior caused by the emergency. Price volatility process is regarded as a
combination of continuous process and jump process. The main conclusions are as
ED
follows:
PT
Firstly, the oil price is characterized by volatility clustering and jump behavior.
As the oil price begins to fluctuate, people will deal in hedging transactions frequently
CE
in the oil futures market to avoid risks, which finally will lead to further wild swings
AC
in the oil price. Oil price will also manifest some jump behaviors under the influence
of unexpected political and economic events. Secondly, oil price shocks have different
effects on agricultural commodities. Cash crops are more vulnerable to the effect of
oil price shocks than food crops. Thirdly, oil price shocks on most agricultural
commodities are asymmetric. For strong wheat, corn, soybean and bean pulp, the
effect of negative oil price shocks is stronger than that of positive oil price shocks,
while the converse is true for natural rubber. Finally, the jump intensity of the oil price
32
ACCEPTED MANUSCRIPT
agricultural commodities, only natural rubber is under influence of the jump intensity
Our results deserve particular attention from policy makers and market investors
P T
in China. First of all, early warning and response mechanisms should be established.
RI
They are effective to cope with the global oil price fluctuations, such as the
SC
diversification of global oil imports and increasing the oil reserves. Secondly, the
NU
effectively resistant to fluctuations in the global oil price. In the end, the kinds of
MA
agricultural commodity futures should be enriched. It is conductive to make the
agricultural commodity futures market become more active. In sum, it is good for
ED
policy makers and market investors to respond effectively to global oil price shocks
PT
Acknowledgments
AC
This research is Supported by the Fundamental Research Funds for the Central
(No.NCET-12-0327).
References
Agiakloglou, C., Newbold, P., 1996. The balance between size and power in Dickey-Fuller tests with
data-dependent rules for the choice of truncation lag. Economics Letters 52, 229-234.
Askari, H., Krichene, N., 2008. Oil price dynamics (2002–2006). Energy Economics 30, 2134-2153.
33
ACCEPTED MANUSCRIPT
Aydın, L., Acar, M., 2011. Economic impact of oil price shocks on the Turkish economy in the coming
Baláž, P., Londarev, A., 2006. Oil and its position in the process of globalization of the world economy.
P T
Beckmann, J., Czudaj, R., 2014. Volatility transmission in agricultural futures markets. Economic
RI
Modelling 36, 541-546.
SC
Econometrics 31, 307-327.
Box, G.E., Jenkins, G.M., 1976. Time series analysis: forecasting and control, revised ed. Holden-Day.
NU
Burke, S.P., 1996. Data-dependent selection of the lag truncation parameter in unit root tests of the
Chan, W.H., Maheu, J.M., 2002. Conditional jump dynamics in stock market returns. Journal of
Chen, S., Kuo, H., Chen, C., 2010. Modeling the relationship between the oil price and global food
Chiou, J., Lee, Y., 2009. Jump dynamics and volatility: Oil and the stock markets. Energy 34, 788-796.
Ciaian, P., Kancs, D.A., 2011. Interdependencies in the energy–bioenergy–food price systems: A
AC
Cunado, J., Perez De Gracia, F., 2014. Oil price shocks and stock market returns: Evidence for some
Du, X., Yu, C.L., Hayes, D.J., 2011. Speculation and volatility spillover in the crude oil and
El-Sharif, I., Brown, D., Burton, B., Nixon, B., Russell, A., 2005. Evidence on the nature and extent of
the relationship between oil prices and equity values in the UK. Energy Economics 27, 819-830.
Ewing, B.T., Malik, F., Ozfidan, O., 2002. Volatility transmission in the oil and natural gas markets.
Fang, C., You, S., 2014. The impact of oil price shocks on the large emerging countries' stock prices:
Evidence from China, India and Russia. International Review of Economics & Finance 29,
34
ACCEPTED MANUSCRIPT
330-338.
Farzanegan, M.R., Markwardt, G., 2009. The effects of oil price shocks on the Iranian economy.
Gronwald, M., 2012. A characterization of oil price behavior—Evidence from jump models. Energy
P T
Economics 34, 1310-1317.
RI
Haixia, W., Shiping, L., 2013. Volatility spillovers in China’s crude oil, corn and fuel ethanol markets.
SC
Hamilton, J.D., 2005. Oil and the Macroeconomy. The New Palgrave Dictionary of Economics
NU
com/dictionary. Jiménez-Rodrίguez, Rebeca and Marcelo Sánchez, 201-228.
Hammoudeh, S., Yuan, Y., 2008. Metal volatility in presence of oil and interest rate shocks. Energy
MA
Economics 30, 606-620.
He, K., Yu, L., Lai, K.K., 2012. Crude oil price analysis and forecasting using wavelet decomposed
ED
Ives, A.R., Abbott, K.C., Ziebarth, N.L., 2010. Analysis of ecological time series with ARMA( p,q )
PT
Iwayemi, A., Fowowe, B., 2011. Impact of oil price shocks on selected macroeconomic variables in
CE
Jimenez-Rodriguez, R., 2008. The impact of oil price shocks: evidence from the industries of six
AC
stationarity against the alternative of a unit root: How sure are we that economic time series have
Lee, Y., Hu, H., Chiou, J., 2010. Jump dynamics with structural breaks for crude oil prices. Energy
Liu, L., 2014. Cross-correlations between crude oil and agricultural commodity markets. Physica A:
Maheu, J.M., McCurdy, T.H., 2004a. News arrival, jump dynamics, and volatility components for
35
ACCEPTED MANUSCRIPT
Maheu, J.M., McCurdy, T.H., 2004b. News arrival, jump dynamics, and volatility components for
Mohammadi, H., 2009. Electricity prices and fuel costs: Long-run relations and short-run dynamics.
P T
Energy Economics 31, 503-509.
RI
Mutuc, M., Pan, S., Hudson, D., 2010. Response of cotton to oil price shocks, The Southern
SC
Narayan, P.K., Narayan, S., Zheng, X., 2010. Gold and oil futures markets: Are markets efficient?
NU
Nazlioglu, S., 2011. World oil and agricultural commodity prices: Evidence from nonlinear causality.
Nazlioglu, S., Soytas, U., 2012. Oil price, agricultural commodity prices, and the dollar: A panel
Ou, B., Zhang, X., Wang, S., 2012. How does China’s macro-economy response to the world crude oil
price shock: A structural dynamic factor model approach. Computers & Industrial Engineering
CE
63, 634-640.
Panagiotidis, T., Rutledge, E., 2007. Oil and gas markets in the UK: Evidence from a cointegrating
AC
Pourahmadi, M., 1992. ARMA Approximations and Representations of a Stationary Time Series.
Rafiq, S., Salim, R., Bloch, H., 2009. Impact of crude oil price volatility on economic activities: An
Scholtens, B., Yurtsever, C., 2012a. Oil price shocks and European industries. Energy Economics 34,
1187-1195.
Scholtens, B., Yurtsever, C., 2012b. Oil price shocks and European industries. Energy Economics 34,
1187-1195.
Soytas, U., Sari, R., Hammoudeh, S., Hacihasanoglu, E., 2009. World oil prices, precious metal prices
36
ACCEPTED MANUSCRIPT
Tang, W., Wu, L., Zhang, Z., 2010. Oil price shocks and their short-and long-term effects on the
Uri, N.D., 1996. Changing crude oil price effects on US agricultural employment. Energy Economics
18, 185-202.
P T
Wang, X., Zhang, C., 2014. The impacts of global oil price shocks on China’s fundamental industries.
RI
Energy Policy 68, 394-402.
Wang, Y., Wu, C., Yang, L., 2014. Oil price shocks and agricultural commodity prices. Energy
SC
Economics 44, 22-35.
Wang, Y.S., Chueh, Y.L., 2013. Dynamic transmission effects between the interest rate, the US dollar,
NU
and gold and crude oil prices. Economic Modelling 30, 792-798.
Yuan, Y., Zhuang, X., Liu, Z., Huang, W., 2014. Analysis of the temporal properties of price shock
MA
sequences in crude oil markets. Physica A: Statistical Mechanics and its Applications 394,
235-246.
ED
Zhang, C., Chen, X., 2011. The impact of global oil price shocks on China’s stock returns: Evidence
Zhang, C., Chen, X., 2014. The impact of global oil price shocks on China’s bulk commodity markets
37
ACCEPTED MANUSCRIPT
Highlights
P T
• The oil price shocks on most agricultural commodities were asymmetric.
RI
• The natural rubber was under the influence of the jump intensity of the oil price.
SC
NU
MA
ED
PT
CE
AC
38