Original Paper Agric. Econ.

– Czech, 60, 2014 (7): 332–342

The linkage between oil and agricultural commodity

prices in the light of the perceived global risk
Giray GOZGOR1, Baris KABLAMACI2
1
Department of Economics and Finance, Dogus University, Istanbul, Turkey
2
Department of Economics, Istanbul University, Istanbul, Turkey

Abstract: The paper examines a systematic interrelationship between the world oil and agricultural commodity prices,
taking the role of the USD and the perceived global market risks into consideration for the period from January 1990 to
June 2013. The authors initially determine the significant cross-sectional dependence in a large balanced panel framework
for 27 commodity prices, and then apply the second generation panel unit root (PUR) tests. Findings from the PUR tests
clearly suggest that there is a strong unit root in agricultural commodity prices. In addition, the empirical findings from the
fixed effects panel data, panel co-integration analysis, the Panel-Wald Causality tests, and the common correlated effects
mean group estimations strongly show that the world oil price and the weak USD have positive impacts on almost all agri-
cultural commodity prices. There are also retained the adjuvant effects of the escalatory perceived global market risk upon
most agricultural commodity prices.

Key words: oil prices, panel data estimations, the VIX

The prices of oil and agricultural commodities have (2009), and von Braun and Torero (2009), have argued
significance for almost all economies. Commodity that in many low-income countries, an abrupt rise
markets came into prominence after the deep boom- in food prices could increase the pervasive poverty,
bust cycle in commodity and oil prices resulting from which would create the economic and political in-
the great global recession that began at the end of stability. In such a crisis, depending on the extent of
2007. Several studies have assessed the interrela- the increase in food prices, these populations may
tionship between the oil and commodity prices and experience irreversible malnutrition in the long run,
some authors have concluded that the relationship is and depending on the extent of the increase in agri-
strong while others stated that it is weak (Baffes 2007; cultural commodity prices, the farmers’ production
Baffes and Haniotis 2010; Pindyck and Rotemberg and marketing costs may significantly increase, and
1990; Plourde and Watkins 1998). In line with the thus, poor net-importing countries may face harsh
previous studies, this paper looks at the strength of challenges.
the relationship. The main objective of this paper is to analyze the
The link between oil prices and the prices of other interrelationship between these two important de-
commodities has been examined by considering sev- cisive factors of the real economic activity: the oil
eral spillover channels. Many researchers have inves- and agricultural commodity prices. Both oil and
tigated the effects of oil prices on the real economic agricultural commodity prices has chiefly gained
activity by analyzing several different transmission prominence in advanced and emerging countries, and
mechanisms, such as the fiscal and monetary policy there are two main explanations for the causal link
channels, which tend to affect the economic growth between the oil prices and agricultural commodity
and welfare (Kilian 2008; Hamilton 2009). Like an oil prices (Headey and Fan 2008). The mechanisms of
price spike, a sharp increase in agricultural commodity macroeconomic performance and commodity price
prices adversely affects economic conditions.1 Some booms can be shaped by fundamental factors, such as
authors, such as Ivanic and Martin (2008), McCalla supply shocks (e.g., hoarding and export restrictions),

1
The causality of prices affecting economic conditions – one must keep in mind that the cause and effect is never in
one direction and mainly depends on a specific context.

332
Agric. Econ. – Czech, 60, 2014 (7): 332–342 Original Paper

weather shocks, productivity slowdowns, stock de- risk and that the index is a strong indicator of the
clines and demand movements (e.g., growth in demand global market conditions. Similarly, Sari et al. (2011)
from China, India, and other emerging countries and used the VIX index to measure the perception of the
biofuel demand). However, non-fundamental factors, global market risk and showed that these percep-
such as the monetary policy stances and futures mar- tions had a notable suppressing effect on oil prices.
kets, which are the determinants of low interest rates, Overall, we suggest that the global risk perceptions
the depreciation of the USD, and financial market can significantly affect the relationship between the
speculations, also affect the pricing mechanisms of oil and agricultural commodity prices.
an economy. Along with these drivers and factors, In addition, we add the real value of the USD to
the regulatory policy changes, such as the passage our empirical models to get more satisfactory results
of the Renewable Fuel Standard of the Energy Policy on the relationship between the prices of oil and
Act of 2005 in the US, have played an important role agricultural commodities. Indeed, a weak USD, the
in the increase of the US ethanol production, which depreciation of the USD against major currencies,
has resulted in a stronger relationship between the leads to higher commodity prices through increas-
oil and agricultural commodity prices and both the ing foreign demand and purchasing power (He et al.
production and demand for biofuels (Zhang et al. 2010). Recent studies also indicate the role of a weak
2010). However, there is no consensus of the effect of dollar on the commodity price inflation that leads to
such policy changes but simply such policy measures increase the commodity prices (Akram 2009; Harri
create an even more complex market situation. et al. 2009).
This paper provides a distinct insight into the The main contributions of this paper to the exist-
examination of the interrelationship between the ing literature are as follows: First, to the best of our
oil and agricultural commodity prices in the light knowledge, this is the first study that uses a second
of the risk perceptions and uncertainty that shape generation panel unit root (henceforth PUR) test by
the global financial market. Indeed, Shewhart (1931) assuming the cross-sectional dependence of panel
distinguishes between the common and special causes. units, namely, agricultural commodity prices, and
He identifies the common causes as the general that uses a common correlated effects mean group
phenomena continuously active within the system panel data estimation technique with a large panel
with a predictable variation. Special causes indicate a framework for agricultural commodity prices. The
new, unexpected, and unpredictable variation within panel data estimation methods have generally a greater
the system, a surprise to its fullest meaning. Thus, statistical power than the time series techniques, due
a distinction is made between uncertainty and risk. to they include information for both time period and
Special causes are also known as the Knightian un- cross-sectional dimension (Nazlioglu and Soytas
certainty and more recently popularized by Taleb 2012). Second, we systematically take into account
as the Black Swan Theory (Taleb 2010). However, the impact of not only the USD but also the VIX index
when speaking of risk perceptions, we complicate the on the relationship between the world oil and agricul-
things further, since perceptions are the subjective tural commodity prices. Third, we firstly use a large
idea of the two by the market participants – affect- balanced panel data framework for 27 agricultural
ing their decisions, which in turn affect the markets. commodity prices over a relatively long time period.
We therefore take into account the indicators of the This is the first paper that to examine direct effects
perceived global risk and global market conditions, of the VIX on the agricultural commodity prices in
namely, the Volatility (VIX) index in a panel data such a large balanced panel data framework. In this
estimation framework. The VIX is the weighted paper we find that the world oil price and the weak
blend of prices for a range of options on the Standard USD have positive impacts on almost all agricultural
and Poor’s (S&P) 500 index – 30 days period and it commodity prices. We also retain the adjuvant effects
indicates the expected movement in the S&P 500 of the escalatory perceived global market risks upon
index over the next 30-day period. The VIX index most agricultural commodity prices.
is an important proxy for the standard deviation of
the S&P 500 returns, where the standard deviation
denotes the average spread of the distribution of re- LITERATURE REVIEW
turns around its mean. For instance, Hartelius et al.
(2008) suggested that the VIX index is a benchmark A growing number of papers in recent years have
proxy for the behaviour of investors in the light of examined the interrelationship between the oil prices

333
Original Paper Agric. Econ. – Czech, 60, 2014 (7): 332–342

and agricultural commodity prices. For instance, Ai commodity markets has resulted in the notable im-
et al. (2006) suggested that the supply side factors portance of the price increases occurring between
affect the price co-movements of wheat, barley, corn, May 2007 and May 2008, which was demonstrated by
oats, and soybeans. Campiche et al. (2007) stated that von Braun and Torero (2009). Nevertheless, Sanders
although there was no co-integration relationship and Irwin (2010) examined the cross-market correla-
between the crude oil prices and the prices of corn, tion between market returns and the positions held
sorghum, sugar, soybeans, soybean oil, and palm oil by the long-only index funds for twelve commodity
for the period from 2003 to 2005, corn and soybean futures markets and showed that the impact of the
prices were co-integrated with the crude oil prices index fund positions on returns across markets was
for the period from 2006 to 2007. Natanelov et al. limited. Capelle-Blancard and Coulibaly (2011) showed
(2011) presented evidence that during the period from the causality between the index investor positions
1993 to 2001, the prices of cocoa, soybeans, soybean and commodity prices on twelve grain, livestock,
oil, wheat, corn and gold were co-integrated with the and other soft commodity markets through a Panel-
crude oil futures prices. However, they observed that Granger causality analysis. Their research indicated
during the period from 2002 to 2010; only the prices that there was no evidence of a causality relationship
of coffee, cocoa, wheat and gold were co-integrated between the index funds and futures prices in the
with the crude oil prices. These studies observe that agricultural futures markets. Byrne et al. (2011) found
the relationship between agricultural commodity and a negative relationship between the real commodity
oil prices is time-specific. prices and the real interest rates and that risk is cap-
Harri et al. (2009) found that corn; cotton, and tured by a measure of the stock market uncertainty.
soybean prices were linked to the oil prices, while On the other hand, He et al. (2010) found that the
the price of wheat was not. The authors also argued real futures prices of crude oil were co-integrated
that the exchange rates are an important factor in with the Kilian economic index, which was used as
the relations among commodity prices over time. an indicator of the global economic activity. These
Gohin and Chantret (2010) presented evidence that researchers also indicated that the trade weighted
the prices of energy and the prices of food could run US Dollar index and the crude oil prices were influ-
in opposite directions when the real income effect enced significantly by the fluctuations of the Kilian
is taken into account. Nazlioglu and Soytas (2012) economic index during both long-run equilibrium
provided strong evidence of the impact of the world conditions and short-run impacts.
oil price changes on most agricultural commodity Finally, Zhang et al. (2010) showed that there was
prices and a positive impact of a weak US Dollar on no direct relation between fuel prices and the agri-
most agricultural commodity prices. These studies cultural commodity prices in the long run. Using the
observe the significant relationship between the copulas framework, Reboredo (2012) showed that the
agricultural commodity and oil prices. agricultural commodity price movements were not
On the other hand, Baffes (2007), Chen et al. (2010), driven by oil price fluctuations.
and Ji and Fan (2012) showed that the impact of the
crude oil market on other commodity markets was
significant when the crude oil prices were at higher DATA AND METHODOLOGY
levels. Using Granger-causality methods, Nazlioglu
(2011) found that the oil and agricultural commodity Data
prices did not cause each other using linear methods
but that nonlinear linkages between these commod- This paper examines a systematic relationship
ity prices exist. In short the relationship between between the world oil price and the agricultural
agricultural commodity and oil prices depend on the commodity prices. This paper also considers the role
specific methodology or the specific market condition. of the USD exchange rate and the perceived global
Some papers investigate the role of speculation and market risks over the period from January 1990 to
uncertainty on the oil and agricultural commodity June 2013. This paper is based on a large balanced
prices nexus. For instance, Gilbert (2010) argued that panel data framework that includes the prices of the
the index-based investment in the agricultural futures 27 agricultural commodities. The frequency of the
markets was the main cause (including macroeco- data used in our study is monthly. This paper focuses
nomic factors) of the recent food price increases. The on the monthly data, as due to such a large number
streaming of speculative capital into the agricultural of commodity prices data are only available at the

334
Agric. Econ. – Czech, 60, 2014 (7): 332–342 Original Paper

monthly frequency. We select the starting date of the Harri et al. (2009), He et al. (2010) and Nazlioglu and
sample based on the availability of the VIX index of Soytas (2012), we use the real effective exchange rate
the Chicago Board Options Exchange (CBOE). We (REER) for the USD. We obtain the REER data from
also report the descriptive summary statistics and the principal global indicators of the IFS. A depre-
the descriptions of all related variables in Table 1. ciation in the USD would cause soaring commodity
We obtain all data on commodity prices and the prices by a channel in rising purchasing power and
world oil prices from the database of the International foreign demand, and thus the effects of the USD on
Financial Statistics (IFS) for commodity prices. To agricultural commodity prices is expected to be nega-
measure the effect of the exchange rate, following tive. Furthermore, following Nazlioglu and Soytas

Table 1. The descriptive summary statistics and the description of variables

Variables Description Unit Average Std. Dev. Skewness Kurtosis

Wheat The United States (the US Golf Ports) USD/t 4.73 0.34 0.72 2.78
Maize The United States (the US Golf Ports) USD/t 4.88 0.38 1.06 2.98
Sorghum The United States (the US Golf Ports) USD/t 4.86 0.35 1.07 3.01
Rice Thailand USD/t 4.48 0.38 0.61 2.62
Barley Canada (Winnepeg) USD/t 4.74 0.37 0.77 2.51
Soybeans United States (Rotterdam) USD/t 4.76 0.35 0.80 2.55
Soybean meal The United States (the US Golf Ports) USD/t 4.74 0.32 0.75 2.56
Soybean oil All Origin (Dutch Ports) USD/t 4.77 0.37 0.65 2.54
Palm oil Malaysia (Rotterdam) USD/t 4.83 0.45 0.31 2.30
Palm kernel oil Malaysia (Rotterdam) USD/t 4.57 0.36 -0.17 2.54
Fishmeal Any Origin (Hamburg) USD/t 4.60 0.46 0.55 2.15
Sunflower oil European Union (European Ports) USD/t 4.16 0.42 0.70 2.79
Olive oil The United Kingdom USD/t 4.19 0.24 0.52 2.22
Groundnuts
Any Origin (Europe) USD/t 4.60 0.37 0.69 2.87
(peanuts) oil
Groundnuts Nigeria USD/t 4.82 0.32 1.02 3.04
Linseed oil Any Origin (World) USD/t 4.15 0.43 0.38 2.11
Beef Australia (the US Ports) US cents/ pound 4.53 0.25 0.60 2.81
Lamb New Zealand (London) US cents/pound 4.41 0.17 -0.27 2.18
Pork The United States US cents/pound 4.56 0.26 -0.30 4.58
Poultry United States (Georgia) US cents/pound 4.49 0.20 0.15 2.04
Sugar Brazil (Free Market) US cents/pound 4.70 0.40 0.42 2.68
Bananas Latin America (the US Ports) USD/t 4.58 0.34 0.06 2.06
Oranges France USD/t 4.26 0.37 0.05 2.02
Copra The Philippines USD/t 4.62 0.43 0.38 3.30
Coffee Brazil (New York) US cents/pound 4.56 0.44 0.02 2.37
Tea Average Auction (The UK) USD/kg 4.64 0.23 0.56 2.34
Tobacco United States (All Markets) USD/kg 4.75 0.17 0.51 2.09
Petroleum
Real World Oil Price USD/barrel 4.13 0.69 0.42 1.77
(crude oil)
Exchange Rate Real Effective Exchange Rate CPI based USD 4.58 0.07 0.49 2.43
Volatility Index S&P 500 VIX Index of the CBOE Level (Monthly) 2.94 0.34 0.46 2.92

All data are in the logarithmic form

335
Original Paper Agric. Econ. – Czech, 60, 2014 (7): 332–342

FAO Food Price Index (2005 = 100) World Crude Oil Price (2005 = 100)
220 300

200
250
180
200
160

140 150

120
100
100
50
80

60 0
90 92 94 96 98 00 02 04 06 08 10 12 90 92 94 96 98 00 02 04 06 08 10 12

Real Effective Exchange Rate of the USD (2005 = 100) VIX

120 70

116
60
112

108 50

104
40
100

96 30

92
20
88

84 10
90 92 94 96 98 00 02 04 06 08 10 12 90 92 94 96 98 00 02 04 06 08 10 12

Figure 1. Graphs of food and oil prices, the real effective exchange rate, and the VIX

(2012), we study with real values and to avoid the that the world oil price changes and the time at which
data inconsistency in commodity prices in different the world oil price affects agricultural commodity
units, our data are based on the price indices (2005 prices, generally by increasing agricultural commod-
= 100) those are obtained from the IFS. We present ity prices when the world oil price rises. We write
the related data in the Figure 1. We use the Food ad down our empirical model in the following equation:
Agriculture Organization (FAO) food price index
(2005 = 100) in Figure 1. lnCOMMOi,t = a0 + a1lnOILi,t–1 – a2lnREERi,t
Following von Braun and Torero (2009), Byrne + a3lnVIXi,t +v1i + v1t + ε1i,1t (1)
et al. (2011) and Sari et al. (2011), we also consider
the impact of the perceived global risk on the world For this equation, lnCOMMOi,t is the price of the
oil price and agricultural commodity prices. For commodity i at time t in logarithmic form, lnOILi,t–1
this purpose, following Sari et al. (2011), we use is the lagged oil price in logarithmic form for cross
a benchmark indicator, namely, the S&P 500 VIX i at time t–1, lnREERi,t is the real effective exchange
index of the CBOE. We obtain the data of the VIX rates of the USD in the logarithmic form for cross i at
from the database of the CBOE, and use the monthly time t, lnVIXi,t is the VIX index in logarithmic form
original data. for cross i at time t, v1i and v1t are cross-section and
period effects, and ε1i,1t is an error term.

Empirical model
Econometric methodology
Following Zhang et al. (2010) and Nazlioglu and
Soytas (2012), among many others, we convert the This paper initially applies the second generation
dependent and explanatory variables into the loga- PUR tests to evaluate the possible persistence in a
rithmic form in the models. We use a lagged world panel framework of agricultural commodity prices.
oil price in the first framework estimations where To the best of our knowledge, this is the first study
agricultural commodity prices are the dependent that uses the second generation PUR tests in a large
variable because there will be a lag between the time panel framework of agricultural commodity prices.

336
Agric. Econ. – Czech, 60, 2014 (7): 332–342 Original Paper

Table 2. Results of the CD test of Pesaran (2004) in agricultural commodity prices (in the logarithmic form)

Cross-sectional dependence Commodity prices (ln)

Pesaran (2004) CD-stat and probability 32.17 (0.000)
Average absolute value of the off-diagonal elements 0.123

Notes: The CD test is defined under the null hypothesis of cross-sectional independence in agricultural commodity
prices; P-value is in parenthesis.

This issue is notably important in overcoming the correlated effects mean group (CCEMG) estimation
shortfalls of the first generation PUR tests that as- technique of Pesaran (2006) to estimate the related
sume a cross-sectional independence by default. For parameters in equation (1) for each panel unit. To
this purpose, we test the cross-sectional dependence the best of our knowledge, this paper is the first that
of 27 agricultural commodity prices for the period uses the CCEMG estimation technique to examine
from January 1990 to June 2013 by using the cross- the relationship between the oil price and agricultural
sectional dependence (CD) test of Pesaran (2004). commodity prices.
Following the results from the CD test of Pesaran The CCEMG estimation technique allows for
(2004), we apply the second generation PUR tests the heterogeneous slope coefficients across panel
to account for the cross-sectional dependence, us- units. This method can successfully eliminate the
ing the methodology proposed by Pesaran (2007). time-variant, unobservable and heterogeneous im-
The results from these PUR tests suggest that there pacts across the panel units as well as the problems
is a strong unit root in the trends of agricultural of identification related to correlation across the
commodity prices. In the light of this finding, we cross-sectionaly dependent panel units. The CCEMG
employ a panel co-integration analysis and panel data technique is also robust despite the presence of a
estimation techniques. Therefore, we focus on the limited number of “strong” factors and an infinite
long-run impacts of the world oil price, the exchange number of “weak” factors, where the latter factors
rate, and the indicator of the perceived global risk can be related to “local market spillover effects” but
(the VIX index) on agricultural commodity prices the former factors indicate ‘global shocks’ (Pesaran
using a panel data estimation framework. and Tosetti 2011). Furthermore, both factors may be
Following the results from PUR tests, we use the non-stationary (Kapetanios et al. 2011). Therefore,
panel co-integration test to determine whether the we initially determine the homogeneity of the long-
long-run relationships exist between the agricultural run parameters by using the Hausman test, and then
commodity and oil prices. We apply the relatively run the estimation technique of Pesaran (2006). We
recent panel co-integration test of Westerlund (2007) report and discuss all empirical findings from these
that allows for multiple structural shifts in series estimation techniques in the next section, which
and takes the cross-sectional dependence among details our empirical findings.
panel units into account. Furthermore, following the
results of the panel data estimations, we check the
robustness of our findings by using the panel-Wald EMPIRICAL FINDINGS
Granger causality tests. Following their promoter
findings to the model in Equation (1), we proceed This section initially reports the findings of the CD
to investigate the validity of our findings from the test of Pesaran (2004), as shown in Table 2.
panel data estimations for each commodity in the As shown in Table 2, the CD test strongly rejects the
long run. For this purpose, we employ the common null hypothesis of the cross-sectional independence.
Table 3. Results of the cross-sectional dependence PUR tests for agricultural commodity prices (in the logarith-
mic form)

Heterogeneous unit root the CIPS (Pesaran 2007) Constant Constant and trend
Zt-bar Statistic –0.847 (0.1723) –0.645 (0.234)

Notes: The CIPS test is defined under null hypothesis of the non-stationary agricultural commodity prices. The CIPS
test assumes cross-sectional dependence that in form of a single unobserved common factor. The optimal number of
lag is selected by the Akaike Information Criterion (AIC). P-values are in parentheses.

337
Original Paper Agric. Econ. – Czech, 60, 2014 (7): 332–342

Table 4. Results of the panel data estimations for agricultural commodity prices (in the logarithmic form)

Regressors (FE) (RE)

***
Lagged Oil Price (ln) 0.239 (0.004) 0.239 (0.004)***
REER (ln) –1.541 (0.042)*** –1.541 (0.041)***
VIX (ln) 0.108 (0.008)*** 0.109 (0.009)***
Observations 7614 7614
2
R (overall) 0.615 0.484
Hausman (robust test) [0.000] –

Notes: Dependent variable is agricultural commodity prices (ln). The constant term is also estimated but is not
reported. We report robust standard errors. Standard errors are in parentheses, and p-values are in brackets.
*** and ** indicate statistical significance at the 1% and 5% levels, respectively. We report p-values of the robust Haus-
man test of Baum et al. (2010) (null hypothesis: random effects estimator is efficient and alternative hypothesis: fixed
effects estimator is consistent).

Accordingly, following the results from the CD test of by 0.24%. Notably, the coefficient that indicates the
Pesaran (2004), we apply the second generation PUR spillover from the oil markets to the agricultural
tests accounting for the cross-sectional dependence, commodity markets was 0.17 in Baffes (2007).
such as the PUR test proposed by Pesaran (2007) and Moreover, a 1% increase in the REER negatively
report the results in Table 3. affects agricultural commodity prices by 1.54%. We
As shown in Table 3, the results from the PUR tests also find a soaring effect of the VIX on the agricul-
of Pesaran (2007) do not reject the null hypothesis tural commodity prices. The coefficient of the VIX
on the non-stationary agricultural commodity prices. is 0.11, which is significant at a 1% statistical level.
In other words, the results from both of the PUR Thus, the perceived global market risk perceptions
tests suggest that there is a strong unit root in 27 are influential on the prices of agricultural commodi-
agricultural commodity prices. In the light of these ties. In addition, we report the panel co-integration
findings, we focus only on the effects of the lagged test of Westerlund (2007) to investigate the long-run
world oil price, the real effective USD exchange rate, relationship between the agricultural commodity and
and the VIX index on the agricultural commodity oil prices in Table 5.
prices that are modelled in Equation (1). The related The results of the robust probability values for
findings are reported in Table 4. four test statistics (Gt, Ga, Pt and Pa) of the panel co-
The results of the robust Hausman test in Table 4 integration test in Westerlund (2007) in Table 5 show
suggest that the fixed effects estimation in column that there is a significant co-integration (long-run
(1) is consistent. The empirical findings in Table 4 relationship) between the agricultural commodity
show that a 1% increase in the lagged world oil price and oil prices. Furthermore, we check the robustness
tends to raise current agricultural commodity prices of our findings using the panel data estimations in
Table 4. For this purpose, we report the findings from
Table 5. Results of the panel co-integration analysis the panel-Wald causality tests in Table 6.
The empirical findings in Table 6 show that there
Westerlund Robust is a unidirectional causality relationship that runs
Value Z-value
(2007) P-value from the world oil price to the agricultural com-
Gt –4.864 –4.203 (0.000) modity prices. On the other hand, the VIX index
Ga –15.73 –3.525 (0.000) also significantly causes agricultural commodity
prices. These findings are in line with the findings
Pt –12.34 –5.263 (0.000)
from the fixed effects panel data estimations shown
Pa –20.65 –8.252 (0.000)
in Table 4. In addition, the REER significantly causes
Notes: Panel co-integration analysis of Westerlund (2007) the agricultural commodity price. This finding is also
has null hypothesis of no co-integration between two ag- consistent with the findings from the fixed effects
ricultural commodity and oil prices. Lag intervals are panel data estimations shown in Table 4, and our
selected by the AIC. main model in Equation (1).

338
Agric. Econ. – Czech, 60, 2014 (7): 332–342 Original Paper

Table 6. Results of the Panel-Wald Causality Tests

Short-run causality (to) Commodity prices (ln) Oil price (ln) REER (ln) VIX (ln)
(from) Commodity prices (ln) – 1.57 [0.2095] 0.26 [0.6097] 1.78 [0.1819]
Oil price (ln) 52.3 [0.0000] – 28.1 [0.0000] 4.97 [0.0268]
REER (ln) 19.8 [0.0000] 4.65 [0.0310] – 13.4 [0.0002]
VIX (ln) 14.1 [0.0002] 85.4 [0.0000] 64.8 [0.0000] –

Notes: The number of lag length is one. P-values are in brackets.

At this point, we also employ the CCEMG estima- of each agricultural commodity. We report the results
tion technique of Pesaran (2006) in order to obtain from the Equation (1) framework by using the VIX
the related parameters in Equation (1) for the price index in the CCEMG estimations in Table 7.

Table 7. Common correlated effects mean group estimation long-run coefficients

Commodity prices (ln) Oil price (ln) REER (ln) VIX (ln)
*** ***
Wheat 0.291 (0.016) –1.884 (0.167) 0.078 (0.033)**
Maize 0.282 (0.021)*** –2.516 (0.173)*** 0.182 (0.035)***
*** ***
Sorghum 0.285 (0.019) –2.072 (0.169) 0.138 (0.036)***
Rice 0.389 (0.020)*** –0.662 (0.174)*** 0.344 (0.042)***
Barley 0.407 (0.019)*** –0.972 (0.178)*** 0.128 (0.034)***
Soybeans 0.259 (0.018)*** –2.336 (0.151)*** 0.193 (0.032)***
*** ***
Soybean meal 0.248 (0.018) –1.778 (0.152) 0.162 (0.035)***
Soybean oil 0.273 (0.019)*** –2.663 (0.161)*** 0.161 (0.031)***
Palm oil 0.290 (0.031)*** –2.679 (0.268)*** 0.233 (0.058)***
Palm kernel oil 0.186 (0.025)*** –2.332 (0.231)*** 0.057 (0.051)
***
Fishmeal 0.534 (0.021) –0.879 (0.178)*** 0.186 (0.040)***
Sunflower oil 0.356 (0.023)*** –2.001 (0.233)*** 0.084 (0.053)
***
Olive oil 0.078 (0.022) –0.010 (0.202) 0.252 (0.237)
*** ***
Groundnuts (peanuts) oil 0.323 (0.020) –1.775 (0.189) 0.165 (0.033)***
*** ***
Groundnuts 0.225 (0.022) –2.045 (0.170) 0.195 (0.035)***
Linseed oil 0.416 (0.023)*** –1.636 (0.227)*** 0.170 (0.046)***
Beef 0.228 (0.015)*** –0.963 (0.136)*** 0.066 (0.051)
*** ***
Lamb 0.113 (0.016) –0.526 (0.113) 0.007 (0.022)
Swine 0.059 (0.025)** –1.289 (0.208)*** 0.071 (0.047)
*** ***
Poultry 0.264 (0.011) –0.290 (0.081) 0.065 (0.021)***
Sugar 0.208 (0.023)*** –3.036 (0.224)*** 0.071 (0.041)*
Bananas 0.316 (0.020)*** –0.911 (0.215)*** 0.172 (0.036)***
Oranges 0.439 (0.018)*** –0.003 (0.215) 0.012 (0.032)
*** ***
Copra 0.254 (0.026) –2.955 (0.249) 0.149 (0.048)***
Coffee 0.089 (0.033)*** –3.677 (0.292)*** 0.143 (0.049)***
Tea 0.207 (0.014)*** –0.523 (0.124)*** 0.203 (0.025)***
Tobacco 0.082 (0.010)*** –1.149 (0.089)*** 0.237 (0.017)***
*** ***
Panel 0.239 (0.005) –1.541 (0.054) 0.108 (0.011)***

Notes: The constant term is also estimated but is not reported. We report robust standard errors. Standard errors are
in parentheses, and p-values are in brackets.
***, ** and * indicate statistical significance at the 1%, 5% and 10% levels, respectively.

339
Original Paper Agric. Econ. – Czech, 60, 2014 (7): 332–342

Table 7 illustrates the effects of the world oil price works; however, the examination of the role of the
on the agricultural commodity prices taking the REER investors’ motivation in the global financial markets
and the VIX index into account. First, our findings on this relationship has remained limited. In this con-
are consistent with the findings of Baffes (2007) and text, this paper attempts to link the global financial
Nazlioglu and Soytas (2012) in that an increase in the markets with the agricultural commodity markets
world oil price has a positive impact on the prices of by considering the channels of uncertainty such as
all agricultural commodities that were examined in the global risk perceptions and the role of the US
this study. Furthermore, the estimated coefficients Dollar. Our empirical findings refer the importance of
are inelastic and they differ from 0.06 to 0.53. Second, fundamental policies in this relationship. The results
the results imply that the changes in the real values show that the oil’s unidirectional positive impact
of the USD have no impact on the prices of olive oil on agricultural commodities is obvious, and given
and oranges. However, real changes in the value of that the crude oil is a key input to the production of
the USD negatively affect the remaining commodity agricultural commodities and agriculture is an en-
prices that were examined in this study, as expected. ergy intensive industry. In addition, policy-induced
In addition, the coefficients of the REER are inelastic diversions of some commodities to the production
only for prices of eight commodities: rice, barley, of biofuels certainly complicate this relationship.
fishmeal, beef, lamb, poultry, bananas, and tea. Third, These results of the paper are in line with those of
the increasing VIX index that shows the increasing the recent studies, such as Baffes (2007), Mitchell
risks perceived by investors in the global financial (2008), Zhang et al. (2010) and Byrne et al. (2011),
markets do not affect the prices of palm kernel oil, which show that the inputs of the ethanol-biodiesel
sunflower oil, olive oil, beef, lamb, swine, and oranges. and biofuel, such as corn, soybean and sugar prices,
We find a positive impact of the VIX index on most are positively correlated with the oil prices. A pos-
of the remaining agricultural commodity prices. The sible rise in the world oil price leads to increases in
significant coefficients for the VIX index differ from the prices of agricultural commodities associated
0.07 to 0.34. with the alternative energy, and thus, the domain
of agricultural commodities increases day by day.
Moreover, Chen et al. (2010) and Reboredo (2012)
CONCLUDING REMARKS AND argued that the co-movement of the agricultural
IMPLICATIONS commodity prices has an important impact on the
portfolio diversification and hedging.
In this paper, we empirically analyze the systematic On the other hand, food-importing, under-devel-
interrelationship between the world oil price and oped and developing countries will be negatively
27 agricultural commodity prices over the period from affected by the rise of the agricultural commod-
January 1990 to June 2013 through a monthly data ity prices. Most of these countries would face not
set. We find a significant cross-sectional dependence only the political and economic instability when
and unit root in a large panel data framework for the agricultural commodity prices rise, but also the
agricultural commodity prices. Using our findings, inevitable adverse effects of nutritional deficiency,
we determine that there are significant impacts of the a reduced capacity to produce nutrient-rich food in
USD and the VIX index on the world oil price and the suitable conditions, and increased poverty. Following
agricultural commodity prices by using the fixed ef- the recent papers of Reboredo (2012) and Ji andFin
fects panel data, the common correlated effects mean (2012), we suggest that financial derivatives might be
group estimations, the panel co-integration-, and the useful tools to decrease the adverse effects of rising
panel-Wald Causality tests. Our findings strongly in- or volatile agricultural commodity prices.
dicate that a weak USD has positive impacts on 25 of In addition, in line with the previous papers by
the 27 agricultural commodity prices. We also retain Gohin and Chantret (2010) and Natanelov et al. (2011),
the adjuvant effects of the escalatory perceived global we find that speculation or, more generally, investor
market risks upon 20 of 27 agricultural commodity motivation is one of the main determinants of the
prices. The soaring world oil price significantly raises agricultural commodity prices. Investors would gener-
all agricultural commodity prices. ally decide to remain in “secure financial positions”;
In recent years, the relationship between the prices namely, they would invest in “palpable goods”, such
of agricultural commodities and oil prices has been as gold, silver, or agricultural commodities, instead
examined by several different methodological frame- of stocks, bonds, or currencies during the periods

340
of financial or economic depression. The empirical Byrne J.P., Fazio G., Fiess N. (2011): Primary Commodity
results in favour of the monumental importance of Prices: Co-movements, Common Factors and Funda-
global financial markets on the market for agricultural mentals. World Bank Policy Research Working Paper,
commodities highlights the need for financial regula- No. 5578.
tion at an international level, particularly during the Campiche J.L., Bryant H.L., Richardson J.W., Outlaw
periods of global recession or during the post-recession J.L. (2007): Examining the evolving correspondence
recovery periods, which can also introduce asset price between petroleum prices and agricultural commodity
bubbles. The monetary policy stances of developed prices. In: American Agricultural Economics Associa-
economies, the global liquidity conditions, and the tion Annual Meeting, Portland, OR, July 29–August
strength of the USD can also be determinants of the 1, 2007.
agricultural commodity prices. However, these sub- Capelle-Blancard G., Coulibaly D. (2011): Index trading and
jects still need to be researched further, particularly agricultural commodity prices: A panel granger causal-
concerning the different impacts of the agricultural ity analysis. Economie Internationale, 126–127: 51–72.
commodity prices on the least-developed or develop- Chen S., Kuo H., Chen C. (2010): Modeling the relationship
ing economies at the national level. between the oil price and global food prices. Applied
In short, we investigate the economic and finan- Energy, 87: 2517–2525.
cial roots of the empirical results and discuss the Gilbert C.L. (2010): How to understand high food prices.
possible implications in the literature. The main Journal of Agricultural Economics, 61: 398–425.
empirical findings in this paper highlight the role Gohin A., Chantret F. (2010): The long-run impact of en-
of the uncertainty and risks perceptions on the ag- ergy prices on world agricultural markets: The role of
ricultural commodity markets and the results are macro-economic linkages. Energy Policy, 38: 333–339.
in line with the recent finance paper of Reboredo Hamilton J.D. (2009): Understanding crude oil prices.
(2012). The remaining results are in line with the Energy Journal, 30: 179–206.
recent economics papers, such as those by Chen Harri A., Nalley L., Hudson D. (2009): The relationship
at al. (2010), Ji and Fan (2012) and Nazlioglu and between oil, exchange rates, and commodity prices.
Soytas (2012). Given the length of the available time Journal of Ag ricultural and Applied Economics, 41:
series, a further research can consider a VAR model 501–510.
for each commodity series and the oil price, where Hartelius K., Kashiwase K., Kodres L. (2008): Emerging
exogenous variables (the VIX and the exchange Market Spread Compression: Is it Real or is it Liquidity?
rates) are considered. Furthermore, this commonly IMF Working Paper No. 08/10.
known econometric methodology could be extended He Y., Wang S., Lai K.K. (2010): Global economic activity
to consider a non-linear relationship between the and crude oil prices: A cointegration analysis. Energy
oil and agricultural commodity prices. Economics, 32: 868–876.
Headey D., Fan S. (2008): Anatomy of a crisis: The causes
and consequences of surging food prices. Agricultural
REFERENCES Economics, 39: 375–391.
Ivanic M., Martin W. (2008): Implications of higher global
Ai C., Chatrath A., Song F. (2006): On the comovement of food prices for poverty in low-income countries. Agri-
commodity prices. American Journal of Agricultural cultural Economics, 36: 405–416.
Economics 88: 574-588. Ji Q., Fan Y. (2012): How does oil price volatility affect
Akram Q.F. (2009): Commodity prices, interest rates and non-energy commodity markets? Applied Energy, 89:
the dollar. Energy Economics, 31: 838–851. 273–280.
Baffes J. (2007): Oil spills on other commodities. Resources Kapetanios G., Pesaran M.H., Yamagata T. (2011): Panels
Policy, 32: 126–134. with non-stationary multifactor error structures. Journal
Baffes J., Haniotis T. (2010): Placing the 2006/08 Com- of Econometrics, 160: 326–348.
modity Price Boom into Perspective. World Bank Policy Kilian L. (2008): The economic effects of energy price
Research Working Paper, No. 5371. shocks. Journal of Economic Literature, 46: 871–909.
Baum C.F., Schaffer M.E., Stillman S. (2010): IVREG29: McCalla A.F. (2009): World food prices: Causes and conse-
Stata module for extended instrumental variables/2SLS quences. Canadian Journal of Agricultural Economics,
and GMM estimation (v9). Statistical Software Com- 57: 23–34.
ponents S4254010. Department of Economics, Boston Mitchell D. (2008): A Note on Rising Food Prices. World
College. Bank Policy Research Working Paper Series. No. 4682.
Original Paper Agric. Econ. – Czech, 60, 2014 (7): 332–342

Natanelov V., Alam M.J., McKenzie A.M., Van Huylen- Reboredo J.C. (2012): Do food and oil prices co-move?
broeck G. (2011): Is there co-movement of agricultural Energy Policy, 49: 456–467.
commodities futures prices and crude oil? Energy Policy, Sanders D.R., Irwin S.C. (2010): A speculative bubble in
39: 4971–4984. commodity futures prices? Cross-sectional evidence.
Nazlioglu S. (2011): World oil and agricultural commodity Agricultural Economics, 41: 25–32.
prices: Evidence from nonlinear causality. Energy Policy, Sari R., Soytas U., Hacihasanoglu E. (2011): Do global risk
39: 2935–2943. perceptions influence world oil prices? Energy Econom-
Nazlioglu S., Soytas U. (2012): Oil price, agricultural com- ics, 33: 515–524.
modity prices, and the dollar: A panel cointegration and Shewhart W.A. (1931): Economic Control of Quality of
causality analysis. Energy Economics, 34: 1098–1104. Manufactured Product. Macmillan, New York.
Pesaran M.H. (2004): General Diagnostic Tests for Cross Taleb N.N. (2010): The Black Swan: The Impact of the
Section Dependence in Panels. IZA Discussion Paper Highly Improbable Fragility. Random House Digital,
Series, No. 1240. Inc., New York.
Pesaran M.H. (2006): Estimation and inference in large Von Braun J., Torero M. (2009): Implementing Physical and
heterogeneous panels with a multifactor error structure. Virtual Food Reserves to Protect the Poor and Prevent
Econometrica, 74: 967–1012. Market Failure. IFPRI Policy Brief 10, Washington DC.
Pesaran M.H. (2007): A simple panel unit root test in the Available at http://www.ifpri.org/sites/default/files/
presence of cross-section dependence. Journal of Ap- publications/bp010.pdf
plied Econometrics, 22: 265–312. Westerlund J. (2007): Testing for error correction in panel
Pesaran M.H., Tosetti E. (2011): Large panels with common data. Oxford Bulletin of Economics and Statistics, 69:
factors and spatial correlation. Journal of Econometrics, 709–748.
161: 182–202. Zhang Z., Lohr L., Escalante C., Wetzstein M. (2010): Food
Pindyck R.S., Rotemberg J.J. (1990): The excess co-move- versus fuel: What do prices tell us? Energy Policy, 38:
ment of commodity prices. Economic Journal, 100: 445–451.
1173–1189.
Plourde A., Watkins G.C. (1998): Crude oil prices between Received: 17th December 2013
1985 and 1994: how volatile in relation to other com- Accepted: 3rd February 2014
modities? Resource and Energy Economics, 20: 245–262.

Contact address:

Giray Gozgor, Dogus University, Department of Economics and Finance, Zeamet Street 21, 34722, Kadikoy, Istanbul,
Turkey
e-mail: [email protected]

342