Tail risk market spillovers of oil to agricultural

commodities: A time-frequency quantile approach

Hammed Abiola Olayinka

Worcester Polytechnic Institute https://orcid.org/0000-0002-9796-5276

Naveed Khan
International Islamic University, Islamabad Pakistan https://orcid.org/0000-0002-4729-040X
Oluwaseun A. Adesina
Ladoke Akintola University of Technology, Ogbomosho, Nigeria https://orcid.org/0000-0001-6952-
3991
OlaOluwa S. Yaya
University of Ibadan https://orcid.org/0000-0002-7554-3507

Research Article

Keywords: Agricultural commodities, Crude oil, Quantile vector autoregression, Time-frequency

connectedness, Network analysis

Posted Date: January 6th, 2025

DOI: https://doi.org/10.21203/rs.3.rs-5743343/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License

Additional Declarations: The authors declare no competing interests.

 Tail risk market spillovers of oil to agricultural commodities: A time-
frequency quantile approach

Abstract
The interplay between commodities and oil prices provides a trade-off for investors and
consumers. However, the increasing interconnectedness between these two series helps
investors to design efficient investment strategies. Therefore, in this paper, the risk spillovers
from oil market prices to that of agricultural commodities, namely, corn, wheat, soybeans,
sugar, cotton, cocoa, coffee, lean hogs, and live cattle for the period spanning from May 1987
to December 2023 was investigated. For the analysis, we employ a quantile frequency
connectedness approach, our findings reveal that under all market conditions (normal, bearish,
and bullish) corn, wheat, and soybean are the return spillovers’ net transmitters, but sugar is a
consistent net shocks receiver across all frequencies. Furthermore, agricultural commodities
receive oil shocks from WTI during normal market conditions. At lower market conditions, the
connectedness between commodities and oil is weaker than that of extreme market conditions
(higher quantiles). Moreover, the network plots show that short-run connectedness dominates
the long-run connectedness at price returns extremes. Similarly, our finding reports significant
implications for policymakers, portfolio managers, and investors to diversify their investments
in different market conditions.
Keywords: Agricultural commodities; Crude oil; Quantile vector autoregression; Time-
frequency connectedness; Network analysis
JEL codes: C22 F21 G11 G13 G32

Conflict of Interest
The authors declare no known conflict of interest that may influence the publication of the paper.

1. Introduction
Crude oil has long been recognised as the most valuable raw material in the world and
an essential source of energy. These days, crude oil is essential for socioeconomic progress
and stability. In the same way as transport vehicles and agricultural machinery used in the
course of agricultural production depend on oil products like gasoline, diesel, and other fuels,
these sources of energy are vital for farming tools (Adam et.al, 2016; Rafiq et al., 2009).
Similarly, increases in price of oil drive up the cost of producing agricultural products because
it affects inputs like transportation, manufacturing equipment, and fertilisers that are heavily
dependent on oils. Furthermore, price increases for these inputs trickle the cost of agricultural
commodities, which drives up the cost of these goods. Therefore, higher prices of oil may
drive the price of agricultural commodity up promptly (Chen, 2010; Vacha et al., 2013).
The current Russia-Ukraine invasion has a direct impact on global agricultural
commodity markets due to sanctions imposed by Russia which led to a cut in the supply of

1
natural gas to some parts of Europe, and recall that natural gas is a fractionated component of
crude oil. A cut in gas supply also affected the pricing of oil and the Organization of Petroleum
Exporting Countries (OPEC) was made to uprise production of oil to meet global demand (Oil
World, 2022; IEA, 2022). Regional policies have a way of influencing energy prices which
have an aftermath effect on the agricultural commodities prices. While the dynamics in the
price of oil at the market may have a triggering influence on the overall energy markets, for
instance, low oil prices would boost oil demand, which results in fewer incentives and interest
from socially responsible investors. In contrast, as the price of oil increases, incentives
increase, which leads to higher investments and profits in agricultural production.
Therefore, it is observed that there exists a nuanced relationship between crop
production and energy sector (Reboredo, 2012; Koirala et al., 2015), and this relationship is
extended to examine the pricing impact of oil substitute products, namely Ethanol and
Biodiesel which is obtained from agriculture products (Shahzad et al., 2018). Considerably,
prices of oil have increased significantly from $30 per barrel to $147 between 2003 and 2008
(Ji et al., 2018). In response to this, the agriculture commodities prices exhibit a similar pattern
as crude oil (Rezitis, 2015; Gong et al., 2023). In recent years, from crude oil to agricultural
commodities, numerous studies have documented unidirectional volatility spillovers (Pal and
Mitra, 2020; Zhang and Qu, 2015), yet some studies document unidirectional and bidirectional
volatility spillovers from agriculture commodities to crude oil (Kang et al., 2019; Naeem et
al., 2022).
Moreover, research on price dynamics of agricultural commodities garner wider
interest of researchers and trading communities. Most of these studies have dealt with the price
return transmission mechanism between oil as a source of energy and agricultural commodities.
For instance, Balcilar, Gabauer and Umar (2021), Furuoka et al. (2023), Tiwari et al. (2018),
Tiwari et al. (2022), and Umar, Jareno and Escribano (2021) have all employed dynamic
connectedness modelling frameworks that include the time and time-frequency Vector
Autoregression (VAR) with consideration for the lead-lag analyses in the oil and food
commodities interdependency and quantile VAR model with the rolling window analyses.
Even though several studies, as fully documented in section 2 (literature review), detailed the
relationships of oil (or energy) with agricultural commodities while other aspects of the price
relations remain scarce. However, no such study which examines the dynamics and pricing
impact of these markets (commodities and oil) in relation to risk spillovers and tail dependency.
Thus, this study is motivated to close the literature gap by investigating risk spillovers and tail
dependence between agricultural commodities and crude oil. This study uses a frequency
2
connectedness approach to assess the potential impact of volatility spillovers from oil markets
to agriculture commodities.
Thus, we significantly add to the emerging literature on the price connectedness of oil
and agricultural commodities by employing the time-frequency dynamic connectedness
method, that is, the quantile frequency VAR model developed by Chatziantoniou et al. (2021).
Specifically, we investigate the interconnectedness between oil markets’ prices and prices of
prominent food commodity with emphasis on how oil influences commodities, and how
commodities influence oil, at the short-term, define between one and five days, and long-term,
defines as more than five days connectivity. Net transmitter of shocks is the market that
influences the other, while net receiver of shock is the market that receives the impact of
shocks. The attractiveness of the quantile frequency VAR connectedness method is that the
method permits for returns connectedness at various quantile values and time frequencies to be
investigated in a unified framework, while the approach of Chatziantoniou, Gabauer and
Stenfors (2021) is the case of quantile time connectedness that checks returns connectedness
at only one frequency point for various quantiles. In short, this study answers the following
questions; (i) Is any increased connectedness between agricultural commodities and oil over
time resulting from an increased global economic uncertainty?1 (ii) How does connectedness
behave during lower and upper quantile point and at the median, that is, the market conditions
where price returns are at extreme large negative, and at extreme large positive, as well as in
the normal market where there is calmness in pricing? (iii) How do oil and commodities behave
in short-term, and in long-term horizons (as defined earlier)? To what extent does energy
dominate agricultural markets? The analyses here will go a long way to influence the decision
of policymakers and oil and commodity traders in strategizing policies for the diversification
of assets.
To address these questions, the price returns of both West Texas Intermediate (WTI)
and Brent oil prices were used. These are proxies for energy, but it is more convenient to use
the Brent oil and WTI markets due to differences in the oil market prices. Agricultural
commodities considered are wheat, soyabeans, corn, cotton, coffee, sugar, cocoa, lean hogs,
and live cattle. Our work renders some thrilling contributions. (i) It involves nine agricultural
commodities in their connectedness with oil price dynamics at the two recognized global oil

1
More recent economic turbulences that cause uncertainty are the September 11 terrorist attack in the US, the
2008/2009 global financial crisis, the Gulf War II, the European sovereign debt crisis, the Brexit case in the UK,
the trade wars, the Corona-Virus pandemic, and the ongoing Russia-Ukraine and Israel-Palestine wars, among
others and these affect oil price at markets inform of oil shocks (Yaya, Abu and Ogundunmade, 2021; Yaya et al.,
2024). Russian-Ukraine war has influenced the production and transportation of agricultural commodities as a
result of decisions on safe corridors for civilians.
3
markets, Brent and WTI. (ii) It analyses quantile connectedness between oil and agricultural
commodities in order to investigate how energy pricing at the two oil markets influences the
listed commodities during the extremely high positive and extremely low negative returns, and
at normal returns when both agricultural commodities and oil markets interact. Similarly,
Londono (2019) noted that the extreme lower end, corresponding to 5% quantile, and extreme
upper end, corresponding 95% quantile, harbour much information regarding both positive and
negative news. Thus, in this regard, using a mean or median-based dynamic connectivity metric
in analysing dynamic connectedness under study may produce inaccurate results. Therefore,
the tail behaviour in the agricultural commodities and oil price returns is explained by the tail
connectedness and spillovers structure of returns from the markets, which is crucial information
for investment and policy decisions. (iii) Frequency connectedness is studied under extreme
market conditions since shocks in various phases of the oil and commodity markets may have
different effects on investment decisions over different time horizons. (iv) Our empirical results
indicate heterogeneity in the frequency connectedness across the tail structure of returns and at
the median value. Pricing at oil markets (WTI and Brent) have little impact on agricultural
commodity markets, at both short-term and long term. Also, the role of corn, soybean and
wheat as highly traded food commodities is shown, as the three dominate oil and other
commodities at the two horizons connectedness for the tail and median risk analyses, while at
the normal market, the connectedness between agricultural commodities and oil, and among
agricultural commodities is weak and disentangled in the case of short-run connectedness.
Following this, our study reports significant implications for portfolio managers and
policymakers to broaden their investment in extreme (lower and higher) market conditions.
The outline of the remainder of the paper is as follows. In Section 2, previous works on
the connectedness of oil and agricultural commodities are reviewed. In section 3, the quantile
frequency VAR econometric method is detailed with the connectedness measures. This section
further presents the dataset used with some preliminary analyses. Section 4 presents the
empirical findings, and Section 5 concludes the paper.

2. Review of Literature
The importance of crude oil a major energy source that drives commodity prices as well as
services has led to studying co-movement and the dynamic linkages between agricultural
commodities and oil. For instance, Lu et al. (2019) examined the dynamics of spillovers
between agricultural commodities and oil prices using a dataset obtained after the global
financial crisis (GFC). Their study employs a bivariate VAR technique of Diebold and Yilmaz
4
(2012) to analyse short-term spillovers which are bidirectional, while in the medium and long-
term, corn price returns is spilled over to oil price returns. Further results show that non-energy
and energy indicate patterns of lower integration after the GFC. Similarly, Ahmadi, Behmiri,
and Manera (2016) investigate the influence of shocks from prices of crude oil on agricultural
commodity price returns. Their empirical analysis uses impulse response functions which is
depends on structural VAR methodology. The authors discover that the response of commodity
price returns spillover to various oil shocks varies. Additionally, they assert that, compared to
the time before and after the crisis, the consequences of oil shocks are significantly stronger
during GFCs. The association between oil returns and agricultural commodities is examined in
Koirala et al. (2015), where a high association between energy prices and non-energy
commodities is observed. Energy-intensive agricultural commodities are also observed to drive
the co-existence of energy and agricultural commodities.
Tiwari et al. (2018) analyze the time-frequency co-movement and lead-lag relationship
between 21 agricultural commodities and oil price indices. They found a high level of co-
movement in long-term between oil and cotton, coal, fishmeal, rice, maize, wheat, and rubber,
while short-term investors are advised to invest in pork and beef markets. Serletis and Xu
(2019) examine the volatility spillovers and mean between the primary feedstock biofuel
markets such as soyabean, corn, and sugar and crude oil market. The authors employ the
dummy variable in the equations, and the potential impacts of the ethanol requirement are
investigated. Specifically, a four-variables VEC–GARCH–in mean model using a BEKK
representation for the variance equation is employed to obtain a strong link between biofuel
feedstock and oil markets.
Dahl, Oglend and Yahya (2020) study the spillover effects between 10 major
agricultural commodities and crude oil by employing Diebold and Yılmaz (2012) spillover
frameworks. The analysis indicates that during the post-2006 period, crude oil emerged as
information’s net receiver after dividing the data into two periods: pre- and post-2006.
Additionally, the study discovers a bidirectional and asymmetric information flow between
crude oil and agricultural commodities, which is more intense during times of economic
uncertainty. Guhathakurta et al. (2020) study the effects of oil price shocks on metal and agro
prices. They recommend optimal portfolios for each period. They identify structural breaks and
use network analysis to understand shocks transfer. Hau et al. (2020) examine the relationship
between global crude oil and Chinese agricultural futures volatility using a quantile-on-quantile
approach. For various quantiles, they discover heterogeneous relationship between the crude
oil volatility and the Chinese agriculture futures volatility. The study shows that low or extreme
5
high quantiles of oil returns have a noticeable influence, but crude oil volatility does not affect
the volatility of agricultural in the usual mode of the crude oil market. The study of Živkov,
Manić, and Đurašković (2020) investigates the long-term and short-term spillover transmission
from oil to food commodities such as canola, wheat, soybean, and maize. Using the component
GARCH model, the authors create both permanent and transient volatilities while taking six
distinct distribution functions into account. In the robust quantile regression approach, the
generated volatility time series are embedded. The oil market's transitory effect has a
marginally high effect on agricultural commodities compared to its permanent counterpart,
indicating that basic factors are not as strongly influenced by short-term information flow.
Based on the data, soybean futures are the best diversification tool when combined with oil
since they are least vulnerable to shocks related to oil volatility. Maitra, Guhathakurta and Kang
(2021) study the asymmetric volatility spillover between commodity and oil markets. They
recommend optimal portfolios for pairs of oil-commodity with better hedge effectiveness,
which are of importance to investors and decision makers. Sun et al. (2021) discovered a two-
way causal relationship between the price of agricultural commodities and crude oil. Contrary
to most studies, their results revealed that oil prices was affected by agricultural commodity
prices as much as vice versa. The study also finds that both markets remained immune to the
shocks during the corona-virus pandemic.
Umar, Jareño and Escribano (2021) study the link between oil price shocks and
agricultural commodities using Antonakakis and Gabauer's (2017) time-varying parameter
vector autoregression (TVP-VAR) technique. Oil risk shocks exhibit stronger directional return
and volatility connectivity than oil demand and supply shocks, according to Umar et al. (2021)
research on the dot-com bubble, the Global Financial Crisis (GFC), and the coronavirus
pandemic outbreak. During the coronavirus pandemic and GFC crisis, there was a surge in the
net return connectivity between receivers, orange juice, lean hog, sugar, and rubber, and
transmitters, canola and corn; however, the net volatility connection was less evident. An
enhanced joint connectedness method based on time-varying parameter vector autoregression
(TVP-VAR) is developed by Balcilar, Gabauer, and Umar (2021) to examine the
connectedness of 11 agricultural commodities and oil. The findings indicate that there is
heterogeneity and that economic events drive the system-wide connection, with maxima
occurring during the coronavirus epidemic, the European Governmental Debt Crisis, and the
GFC. The analysis also shows that, in most of these markets, crude oil is equally susceptible to
innovations and acts as the primary net transmitter of shocks, impacting other commodity
markets. Hung (2021) studied the effects of crude oil prices on agricultural commodity markets
6
during the coronavirus outbreak. The study finds that there is a more significant return spillover
during the crisis, and the intensity of the relationship varies over time. Important variations are
also found in the amount of spillover to crude oil prices amongst agricultural commodity
markets. Naeem et al. (2022) find that oil shocks and agricultural commodities have strong
intra-connectedness but weaker inter-connectedness.
Tiwari et al. (2022) employed a QVAR model to investigate the volatility spillover
among agricultural commodity, energy, and biofuel markets. They find that connectedness is
stronger during extreme market movements with volatility spillovers from agricultural to
energy markets are significant. The degree of connectedness and volatility spillovers has also
been altered by the COVID-19 era. The study has practical ramifications for risk mitigation for
investment portfolios. Yoon (2022) found that there is a substantial short-run bidirectional
causation among the returns of ethanol, WTI oil, and maize prices, but no cointegration based
on quantiles of these three commodities when all quantiles are included. The transmission risk
between WTI crude oil prices and many significant commodities markets is examined by
Boroumand and Porcher (2023). Between 1982 and 2020, the authors assess the volatility
contagion under different market situations and commodity price cycles and further investigate
the dynamics of commodity connection. Their findings demonstrate that macroeconomic
factors affect the spread of contagion's severity, speed, and duration throughout different
periods.
Cui and Maghyereh (2023) study the risk connectedness between oil and commodity
futures using a time-varying parameter-vector auto-regression-based technique. They analyze
the correlation between various futures contracts and construct hedging and optimal-weighted
portfolio strategies. Findings indicate that significant crises have strengthened the connections
between global oil and commodities futures and that the optimal-weighted portfolio reduces
risk more than hedging does. Additionally, the portfolio with the lowest kurtosis connectivity
has the highest cumulative returns. Montasser, Belhoula, and Charfeddine (2023)’s findings
show that, for various lags, there are significant shocks transmitting effects from Brent to
agricultural commodities; only the commodities corn and soybeans exhibit a reverse leading
effect. Furuoka et al. (2023) apply the TVP-VAR model in examining the short- and long-term
relationships between agricultural and energy commodities. They discovered that the
coronavirus pandemic caused the connectivity index to fall below 40% in 2020–2021, from
approximately 40–60% in 2018–2019. The study also discovered that there is a noticeable
asymmetry in the network of connection and that wheat, corn, and flour transmit risks to natural

7
gas and oil. Portfolio analysis shows that to maximize profits, investors need a low percentage
of energy in their energy-agricultural commodities portfolio.
Wang et al. (2023) employ quantile connectedness to investigate the dynamic spillover
among agricultural product prices, crude oil prices, and climate policy uncertainty, Findings
revealed that extreme quantile has a higher degree of connectivity. System connectivity is
impacted by uncertain climate policy initiatives and significant occurrences. Additional
analysis reveals that climate policy uncertainties in the extreme market, lower and upper
quantile, was the continuous shock receivers in the system, in contrast to their isolated role in
the typical market (median quantile). Wu et al. (2023) compare the volatility of connectivity
brought on by the coronavirus pandemic, and the ongoing Russia-Ukraine war using the TVP-
VAR frequency connectedness technique. Increased connectivity is brought about by both
shocks, but the pandemic shock is more potent. Early into the pandemic, high- and medium-
frequency connectivity predominate, whereas low-frequency connectivity is more prevalent
during hostilities. Furthermore, at the early stages of the pandemic, fossil fuels are the risk
transmitter, whereas, during the conflict, agricultural commodities take on that role. To avoid
inflation and an economic downturn, we need to take precautions against the hazards of
agricultural commodities and fossil energy spreading to the post-conflict economy.
From a methodological perspective, we make a distinct contribution to the expanding
corpus of literature on the dynamic interdependence between the agricultural commodities and
oil prices. Specifically, we depart from the earlier Chatziantoniou, Gabauer and Stenfors (2021)
model by employing an updated quantile-based method that takes frequency domain
connectedness into account, as done by Baruník and Křehlík (2018). The utilized methodology
is predicated on the idea that the degree of connectivity can fluctuate on the account of the state
of the market (bullish and bearish, as well as typical market circumstances) and between
frequencies.

3. Methodology
The effect of structural shocks on each network variable's volatility is explained by the forecast
error variance decompositions (FEVD). To put it briefly, significant co-movements between
the network's variables are mirrored upon substantial error variances. For instance, quantile
connectedness is a method used to determine whether there is a differing strong co-movement
among the variables at both positive returns’ extreme and negative returns’ extreme, that is,
both extreme positive structural shocks, which is the case of higher quantiles, and extreme
negative structural shocks, which is for lower quantiles. It is worthy of note that frequency
connection is a recent advancement that was introduced by Barunik and Krehlik (2018). It
8
separates connectedness into two categories: low-frequency connectedness (a connectedness
type where shocks lead to structural changes within the variable network due to the long impact
shocks have) and high-frequency connectedness (a connectedness type where shocks last for a
short period within the network of variables, and thus have a short impact). This attractiveness
rendered by the frequency connectedness framework is hardly employed and is of policy
relevance.

We start by presenting the modelling framework for the quantile frequency VAR of
Chatziantoniou et al. (2021) by presenting first the quantile time VAR framework of
Chatziantoniou, Gabauer and Stenfors (2021). A quantile-specified QVAR model of order 𝑝
is, 𝑦𝑡 = 𝜇𝑡,𝜏 + Φ1,𝜏 𝑦𝑡−1 + Φ2,𝜏 𝑦𝑡−2 + ⋯ + Φ𝑝,𝜏 𝑦𝑡−𝑝 + 𝜖𝑡,𝜏 (1)

where 𝑦𝑡 and 𝑦𝑡−𝑖 (𝑖 = 1, … , 𝑝) are 𝑁 × 1 dimensional endogenous variable in vector spaces,

𝜏 represents a quantile value set as [0,1], 𝑝 represents the lag of the autoregression in the QVAR
model, 𝜇𝑡,𝜏 is an 𝑁 × 1 dimensional conditional mean vector for a quantile 𝜏, Φ𝑖,𝜏 is an 𝑁 × 𝑁
dimensional coefficient matrix for the QVAR model, and 𝜖𝑡,𝜏 is the 𝑁 × 1 dimensional error
′
vector, such that variance-covariance matrix, Σ𝜏 = 𝜖𝑡,𝜏 𝜖𝑡,𝜏 has an 𝑁 × 𝑁 dimension. As part of
the derivation, the QVAR(p) model is simplified to its quantile vector moving average
representation, QVMA(∞) as,

𝑦𝑡 = 𝜇𝑡,𝜏 + Ψ0,𝜏 𝜖𝑡 + Ψ1,𝜏 𝜖𝑡−1 + Ψ2,𝜏 𝜖𝑡−2 + ⋯ (2)

Using Wold’s interpretation. The generalized forecast error variance decomposition (GFEVD)
of Koop et al. (1996) and Pesaran and Shin (1998) is then computed from eq (2). Similarly,
Diebold and Yilmaz (2012, 2014) is a mean-based connectedness approach, while the dynamic
connectedness such as the QVAR being discussed is variance-based, relying on GFEVD. Thus,
the GFEVD is interpreted as the impact of a shock from a variable 𝑗 on other variables 𝑖 in
terms of its share of forecast error variance. The variance share is obtained as,
2
−1 ∑𝐻
Σ𝜏,𝑗,𝑗 ℎ=0[(Ψℎ,𝜏 Σ𝜏 )𝑖,𝑗 ]
𝜃𝑖,𝑗,𝐻 = ∑𝐻 ′ (3)
ℎ=0(Ψℎ,𝜏 Σ𝜏 Ψℎ,𝜏 ) 𝑖,𝑖

which is normalised as,

𝜃
𝜃̃𝑖,𝑗,𝐻 = ∑𝑁 𝑖,𝑗,𝐻
𝜃
(4)
𝑗=1 𝑖,𝑗,𝐻

due to non-summable to unity the rows of 𝜃𝑖,𝑗,𝐻 such that ∑𝑁 ̃ 𝑁 𝑁 ̃

𝑗=1 𝜃𝑖,𝑗,𝐻 = 1 and ∑𝑗=1 ∑𝑖=1 𝜃𝑖,𝑗,𝐻 =

𝑁. Thus, summing to 1 of each row represents the total impact a shock in variable 𝑗 has
9
influenced itself plus its influence on all other variables 𝑖. That describes the connectedness for
time domain which does not cater for the splitting of impact of connectedness into short-term
and long-term.

For the frequency domain, we dwell on the spectral decomposition approach detailed
in Stiassny (1996), which uses the frequency response function Ψ(𝑒 −𝑖𝜔 ) = ∑∞
ℎ=0 𝑒
−𝑖𝜔ℎ
Ψℎ
where 𝑖 = √−1 (a complex number), and 𝜔 is the spectral density frequency for 𝑦𝑡 , that is the
Fourier representation of the QVMA(∞) in eq (2), given as,

𝑓𝑦,𝜔 = ∑∞ ′
ℎ=−∞ 𝐸(𝑦𝑡 𝑦𝑡−ℎ )𝑒
−𝑖𝜔ℎ
= Ψ(𝑒 −𝑖𝜔ℎ )Σ𝑡 Ψ′ (𝑒 −𝑖𝜔ℎ ) (5)

The frequency GFEVD is then seen as the combination of the spectral density function (9) and
the time domain GFEVD in eq (3). Thus,
2
Σ−1 ∞
𝜏,𝑗,𝑗 |∑ℎ=0[Ψℎ,𝜏 (𝑒
−𝑖𝜔ℎ )Σ ] |
𝜏 𝑖,𝑗
𝜃𝑖,𝑗,𝜔 = ∑∞ −𝑖𝜔ℎ )Σ Ψ′ (𝑒 −𝑖𝜔ℎ )]
(6)
ℎ=0[Ψℎ,𝜏 (𝑒 𝜏 ℎ,𝜏 𝑖,𝑖

which is also normalised as,

𝜃
𝜃̃𝑖,𝑗,𝜔 = ∑𝑁 𝑖,𝑗,𝜔
𝜃
(7)
𝑗=1 𝑖,𝑗,𝜔

where 𝜃̃𝑖,𝑗,𝜔 is a point in the spectrum of the ith series for a frequency 𝜔 which is meant for a
shock in the jth series. This is a within-frequency measure used to deduce the short-run and
long-run connectedness, which is deemed better than the connectedness at a single frequency.
All frequencies available in the connectedness are first assumed to be within the interval, 𝑑 =
(𝛼, 𝛽) such that 𝛼, 𝛽 ∈ (−𝜋, 𝜋), 𝛼 < 𝛽, and these are aggregated based on the integration,

𝛽
𝜃̃𝑖,𝑗,𝑑 = ∫𝛼 𝜃̃𝑖,𝑗,𝜔 𝑑𝜔 (8)

The frequency connectedness measures in a certain frequency range 𝑑 can then be obtained in
a similar manner to Diebold and Yilmaz (2012, 2014).

In the frequency connectedness, there are two frequency ranges, implying short-term
𝜋 𝜋
and long-term, that is, 1 to 5 days [𝑑1 = ( , 𝜋)] and from 6 to infinite days [𝑑2 = (0, )]. The
5 5

short-term new pairwise directional connectedness (NPDC) is then computed as,

𝑁𝑃𝐷𝐶𝑖,𝑗,𝑑1 = 𝜃̃𝑖,𝑗,𝑑1 − 𝜃̃𝑗,𝑖,𝑑1 (9)

that is the net impact in the two variables paired together, whereas in a case with 𝑁𝑃𝐷𝐶𝑖,𝑗,𝑑1 >
0 implies the dominance of the first variable 𝑗 over the other variables 𝑖 at that particular
10
frequency. The short-term net total directional connectedness is the difference between the
short-term total connectedness from 𝑗 to others (other variables 𝑖), and the short-term total
directional connectedness from others (variables 𝑖) to 𝑗. This is obtained as,

𝑁𝐸𝑇𝑗,𝑑1 = 𝑇𝑂𝑗,𝑑1 − 𝐹𝑅𝑂𝑀𝑗,𝑑1

= ∑𝑁 ̃ 𝑁 ̃
𝑗=1,𝑖≠𝑗 𝜃𝑖,𝑗,𝑑1 − ∑𝑗=1,𝑖≠𝑗 𝜃𝑗,𝑖,𝑑1 (10)

and if 𝑁𝐸𝑇𝑗,𝑑1 > 0, it implies that variable 𝑗 is a net shock transmitter to all other variables 𝑖
in the network at the short-term connectedness. The NPDC is the case of a one-to-one variable
assessment, while the case of NET is a one-to-many relationship and these should not be
confused together. Lastly, the total connectedness index (TCI) for a short-term total
connectedness gives the strength of connectedness of all variables in the network and a high
TCI implies a high market risk and vice-versa. Similarly, the long-term connectedness
measures 𝑁𝑃𝐷𝐶𝑖,𝑗,𝑑2 , 𝑇𝑂𝑗,𝑑2 , 𝐹𝑅𝑂𝑀𝑗,𝑑2 , 𝑁𝐸𝑇𝑗,𝑑2 , and 𝑇𝐶𝐼𝑑2 are obtained correspondingly.

3.1 Data
The data sample analysed are the daily S&P GSCI indices for agricultural commodities,
spanning from May 1987 to December 2023. The agricultural commodities include corn,
cotton, cocoa, coffee, lean hogs, live cattle, soybeans, sugar, and wheat. The commodities data
were retrieved from Datastream. Oil prices are those from WTI market in the United States,
and European Brent market, and the data were retrieved from the online database of US Energy
Information Administration. Figure 1 displays price plots for various commodities, showing
that WTI and Brent oil have comparable price patterns. Oil has witnessed several price crashes
of which significant ones are that of GFC in 2007-2008, the 2015 crash due to the
overproduction of oil at markets, the coronavirus pandemic, and the 2022 crash due to on
Russia-Ukraine invasion. By looking at the historical price indices of agricultural commodities
plots, no consistent plotting pattern is observed among the nine series, and with oil price series.
Price indices are nonstationary and transformed stationary equivalence is required. This is
achieved by carrying out percentage changes of each series, which is used as a proxy for the
return series. The return series are then checked against heteroscedasticity. Further, in Table 1,
the mean returns for Brent, WTI, soybeans, sugar, and live cattle are positive implying that
pricing of these commodities increases more than it decreases. Negative mean returns are
recorded for wheat, corn, coffee, cocoa, and lean hogs implying that prices of the commodities
have more tendency to decrease. Estimates of variance indicate that Brent and WTI are more
11
volatile compared to agricultural commodities while live cattle are the least volatile. Moreover,
return series are negatively skewed in most cases, and all the series have been found to exhibit
significant leptokurtosis through the excess kurtosis estimates. Thus, Jarque-Bera (JB) test
statistics have been decisively rejected in all cases, indicating non-normality. After doing the
ERS unit root test as described in Elliot et al. (1996), the null hypotheses are consistently and
resoundingly rejected, which is expected as returns series are often stationary. Furthermore, the
Q test for serial autocorrelation and the Q2 tests for ARCH/GARCH errors have shown
rejections of no serial correlation, which predicates the possibility of modelling such serially
correlation returns with heteroscedastic-based model.
=======FIGURE 1=======
In the middle panel of Table 1, Pearson correlation test results are presented and
significant positive correlations are found between price returns of oil and commodities.
Correlations are also found among agricultural commodities. The correlation test is often
challenged since it is designed for testing linearly-supposed dependent time series, while oil
and agricultural commodities are financial series that are assumed to be nonlinear due to
structural breaks. Thus, correlation tests alone may not do justice to the interdependence
structure of the return series. The last panel of Table 1, presents the results of the persistence
test on returns series based on variants of Hurst estimates. It is noted in the results that the 11
series persist in a similar way based on their estimates by each estimator and Hurst estimates
are around 0.5 in most cases implying stable/stationary returns series. Further, similar
persistence estimates imply similar cross-sectional dependence in the returns series. This is a
required conclusion before utilizing the QVAR model in checking the dynamic
connectedness between agricultural commodities and oil volatility.
======TABLE 1=======
Estimation of frequency QVAR models for different quantiles, reported here depends
on 200-days rolling-window with lag 1 and 100-step-ahead GFEVD.

4. Empirical Findings
Based on sampled data, results are presented in this section while focusing frequency
connectedness by quantiles (lower, upper, and middle), following Chatziantoniou et al. (2021).
The lower quantile connotes left tail returns risks while the upper quantile connotes left tail
returns risks in the connectedness. Recall, the frequency connectedness approach is more
general as it allows the total connectedness (time connectedness) to be split based on time-
frequency connectedness, at least two (short and long-term), following this we use the short
and long term connectedness framework. Thus, the connectedness approach used here allows
12
to investigate the time-frequency connectedness for extreme positive returns and extreme
negative returns, as well as at normal market conditions where markets are calm.

4.1 Average tail risk and median connectedness

In Table 2, the results of total, short-term (1-5 days) and the long-term (more than 5 days)
connectedness at the lower quantile are presented. It is observed that Brent oil has the highest
share of own-variance spillovers of 13.97% for time connectedness implying that the remaining
86.03% is the spillover contributions received from all other series in the connectedness
network. At the short-term connectedness, Brent has 10.27% own-share variance
corresponding to 3.7% proportion of its own-share variance at the long-term connectedness.
Based on total connectedness, Brent transmits up to 84.7% market spillovers to other assets in
the network while it receives up to 86.03% market spillovers. Thus, Brent oil becomes a net
receivers of volatility shocks (𝑁𝐸𝑇 = −1.33). Based on the short-term connectedness, Brent
transmits 64.19% of market spillovers to others in the network and receives 62.41% making,
Brent a net shock transmitter at the short-term connectedness. During the long-term
connectedness, Brent receives more shocks than the proportion it transmits. It receives 23.62%
market shares of variance and transmits 20.5% share, positioning the oil market as a net shock
receiver at the long-term period. Among agricultural commodities, similar behaviours with
changes in transmission of shocks by frequency is observed for wheat, cotton, coffee, cocoa,
and lean hog. In the case of corn, soybeans, and live cattle, transmission of shocks at the short-
term and long-term do not change while corn and soybeans are net transmitters and live cattle
is a net receiver. WTI oil as a net transmitter of shocks at the total connectedness and short-run
connectedness, white it is a weak net receiver not a receiver of shocks at the long-term
connectedness.
In sum, at the lower quantile, corn and soybeans are net transmitters at the total, short-
and long-term frequency connectedness, while sugar and live cattle are also net transmitters for
three time-frequency cases. The total connectedness index (TCI) is 85.53 and short-run and
long-run TCIs being 64.85 and 20.68, respectively.
======TABLE 2=======
In Table 3, at the upper quantile, which connotes situations whereby returns of oil and
commodities are at extreme high positive values, Brent oil sternly becomes a net shock receiver
for both short-term and long-term connectedness. Similarly, cotton, sugar and coffee follow
the same manner as net receivers irrespective of the time-frequency. Similarly, corn and
soybeans remain net shocks transmitters at the two time-frequency connectedness (short and

13
long-term). Thus, at the extreme high positive and extreme high negative returns, cotton and
soybeans remain strong commodities to withstand in market turbulence, while sugar remain a
net transmitter, even at the upper quantile. There is a pronounced heterogeneity in the
frequency connectedness at the upper and lower quantiles. For example, at the upper quantile,
WTI, wheat, cotton, live cattle and lean hogs relay different decision in their frequency
connectedness. Thus, the time-frequency connectedness employed at the lower and upper
quantile, connoting extreme negative and extreme positive returns is relevant in the context of
this analysis.

======TABLE 3=======
At the median quantile, denoted by 𝜏 = 0.5, we find in Table 4 that own-share variances
have increased compared to the results of lower quantile (see Table 2) and upper quantile (see
Table 3). An increase in own-share variance proportion implies gross reduction in the spillovers
to and from other commodities in the network. For instance, spillovers to and from during
lower quantile are about (82-89)% and (61-68.8)%, respectively, while those at the upper
quantile are around (81-89.57)% and (84.68-85.84)%, respectively. These culminate into high
TCI values. At the normal market, own-share variance proportions are now large as the assets
are less likely to be influenced externally. Thus, FROM and TO spillover proportions are small
leading to very a low average TCI of 31.34 for the total connectedness. It is interesting to note
that only wheat indicates heterogeneity in its frequency connectedness. Brent emerged as a net
shock receiver, and agricultural commodities such as cocoa, coffee, cotton, lean hogs, live
cattle, and sugar, followed suit as net shock receivers. Corn and soybeans are net shock
transmitters at both short- and long-term connectedness.
By comparing the results at the lower, upper, and median quantiles, corn and soybeans
are net shock transmitters at any price return values, implying that the two agricultural
commodities can hedge against others in the network. Sugar is also a consistent net shock
receiver during the normal market condition, denoted by the median quantile frequency
connectedness. Most agricultural commodities are net receivers of shock, mainly receiving oil
shocks from WTI. Heterogeneity in short- and long-term connectedness is more pronounced
during extreme market price returns.
======TABLE 4=======

4.2 Dynamic total and net directional tail risk, and median connectedness

14
At the lower quantile, Figure 2 presents plots of total, short- and long-term connectedness,
where the black coloured is the total connectedness plot and the red coloured and green
coloured are the short run and long-run connectedness plots, respectively, rendered by TCI.
Total connectedness hovers above 80% but below 90% as marked from the plot. Long- term
activities are more pronounced and more volatile in the TCI realizations. In few periods, there
are occasional long spikes (in green) which dominate the short-run (in red), and short- term
frequency generally dominates in the dynamic connectedness at the lower quantile. This
domination is clearly seen in the net total directional connectedness at the lower quantile in
Figure 3. The red spikes shoot out at the positive and negative sides of plots with longer spikes
on many occasions further implying this dominance.

=======FIGURE 2=======
=======FIGURE 3=======
Next, at the upper quantile, short-term connectedness still dominates clearly the long-
term connectedness as in Figure 4 and TCI are above 80%, similar to those in Figure 3. In
Figure 5, the net dynamic connectedness results also indicate the dominance of short-run
connectedness (red part) over the long-run throughout the sample period, as in the case of lower
quantile connectedness. Net directional connectedness plots in Figure 4 for lower quantile and
Figure 5 for upper quantile are weak such that it is difficult to classify if an asset is a net
transmitter or receiver from the plots. Thus, we further make much reliance on the average
results for NET presented in Table 2 and 3, respectively for the two tail risks.
=======FIGURE 4=======
=======FIGURE 5=======
The results, based on the middle quantile are quite different from those reported for
right tail risks (upper 95 quantile) and left tail risks (lower 5th quantile). The total connectedness
in Figure 6 reports low TCI values over the sample period except around 2010 when the TCI
skyrocketed close to 70% and dropped sharply after few months and increased up to 50%
before further drops below 20% around 2002 – 2003. The 2010 drops in TCI is as a result of
global crash in the prices of agricultural commodities (Bellmann and Hepburn, 2017). In 2020,
the TCI spiked up due to COVID-19 pandemic. In this plot, short-term connectedness
dominates throughout without sporadic spiking up of long-term connectedness. This is
expected since the energy and agricultural commodity markets are in state of calmness (normal
market). Apart from the low TCIs reported in the case of median connectedness (50th quantile),
net directional connectedness indicates clear direction of spillover, clearer than those reported
for the tail risks (lower and upper quantiles). Brent, cocoa, cotton, live cattle, wheat, sugar, lean
15
hogs are clearly classified to be net receiver of shocks since a larger part of frequency
connectedness plots are found in the negative NET axes, while soybeans, WTI and corn are net
shock transmitters. Clearly, short-run connectedness dominates the short-run connectedness
here in Figure 7 as well.
=======FIGURE 6=======
=======FIGURE 7=======

4.3 Network plots for tail risk and median connectedness

Net pairwise directional connectedness at the two tail ends of returns as well as the median are
given by network plots in Figures 8-10. Blue coloured nodes are net transmitters of shocks
while orange- coloured nodes are net receivers of shocks. The size of each node tallies with the
magnitudes of net directional connectedness as in Tables 2-4, while the thickness of the joining
line implies the transmission strength.
During the extreme negative returns adjudged by lower quantile, we present in Figure
8 network plots for total (upper left), long-term (upper right), and short-term (lower left)
connectedness. In the three plots, live cattle received most shocks, as it received shocks from
nearly all other commodities in the network. For the short-term connectedness, live cattle
received strongest shocks from wheat compared to what it receives from others. For the long-
term connectedness, soybean transmits fairly more shocks to live cattle than what corn, wheat,
WTI, and Brent transmit. Also, wheat dominates the network as a net transmitter and it
transmits more strongly during the long-term connectedness than during the short-term
connectedness. At the upper quantile, wheat only transmits more strongly to cocoa, and
transmits weakly to WTI and Brent at the long-term (see Figure 9). At the short-term, wheat
becomes a net receiver of shocks, while WTI transmits strongly to Brent. Clearly, the role of
frequency connectedness in the oil-agricultural commodity connectedness is shown in the two
tails risk analysis. Lastly, at the median quantile, the connectedness is weak, and this has been
indicated by low average TCI value (see in Table 4). During the normal market in the long-
term connectedness, WTI transmits shocks to sugar and its oil counterpart, brent. Soybean and
corn being strong parties in the connectedness transmits shocks weakly to agricultural
commodities (see Figure 10), while at the short-term connectedness for the normal market
condition, transmission of shocks from oil to agricultural commodities, or within agricultural
commodities, are disentangled as a result of weak average TCI in Table 4. Though, the net
directional connectedness in Table 4 gives clear directions of connectedness but the fact that
the joint connectedness is weak leads to the disentanglement in the network plot in Figure 10.

16
In the overall, the total connectedness indicates the transmission of shocks from WTI to Brent,
while corn and soybeans transmit weak shocks between themselves, and corn further transmits
shocks to wheat.
=========FIGURE 8========

=========FIGURE 9========

=========FIGURE 10========

5. Conclusion
The financialization of commodities has resulted in a significant level of unpredictability in the
crude oil and agricultural markets. Commodity price fluctuations have also been greatly
influenced by geopolitical conflicts, economic crises, and energy-related issues. In the
agricultural sector, oil is one of the most vital inputs. Costs associated with agriculture rise
when oil prices rise. Therefore, the association between crude oil and commodity futures
requires a deeper comprehension and measurement of the spillover effect and dependence
structure of these series. Furthermore, the role of energy in agricultural production and
transportation of produce is of paramount importance. To do this, in this paper, we investigate
the link between oil markets and agricultural commodity prices, namely, wheat, corn, soybeans,
cotton, sugar, coffee, cocoa, live cattle, and lean hogs, with emphasis on how oil influences
these agricultural commodities, and how commodities influence oil, at the short-term and long-
term time-frequency windows. The daily price returns of WTI oil and Brent oil prices, as well
as S&P GSCI indices of agricultural commodities, spanning from May 1987 to December
2023, are analysed by employing the quantile frequency VAR model of Chatziantoniou et al.
(2021), of which the econometric method has special attractiveness in analysing multiple
variables by time-frequency and quantile in a unified framework. The model proposed in
Chatziantoniou et al. (2021) allows for quantile connectedness to be investigated at only one
frequency point which is not the case in the interaction of economic and financial returns. It is
noted that during the interactions and spillovers, some shocks are transmitted at the long run
(above 5 days) while others at the short run (1-5 days).
At the lower quantile only, sugar and live cattle emerged as net shock receivers by
frequency, while WTI, sugar, coffee, and cocoa are net shock receivers at the upper quantile of
returns. Thus, for both lower and upper quantiles, corn and soybeans are notable net
transmitters of shocks for the two time frequencies, while sugar is a consistent net shocks
receiver during three time-frequency cases. Pronounced heterogeneities in short- and long-term
connectedness is obvious during extreme market price returns, for instance, at the upper
17
quantile, WTI, wheat, cotton, live cattle and lean hogs relay different decisions in their
frequency connectedness. At the median quantile, WTI, corn, and soybean are net transmitters,
while Brent and the remaining agricultural commodities indicate net shock receiving
tendencies by frequency except wheat which indicates mixed results in its frequency
connectedness. Thus, at the lower, upper, and median quantiles, corn and soybeans stand out
as net shock transmitters at any price return values for long-run and short-run, implying their
tendencies of being used as a hedge against others in the network, while sugar remains a net
shock receiver. Further, during the normal market situation (median quantile), agricultural
commodities are seen to receive oil shocks from WTI.
Similarly, our findings report manifold implications for policymakers, portfolio
managers, and agriculture investors. From the perspective of investors, to maximize the
benefits of diversification, and because of the asymmetric reliance, asset allocation and
dynamic portfolio management are crucial between oil and agriculture commodities
prices when market conditions are at the extreme. Moreover, the existence of agricultural
commodities' contingent connections necessitates a deeper comprehension of how they provide
hedging effects against uncertainty in the oil markets and enable the accurate futures contract
pricing throughout a range of market situations. Furthermore, the dominance of short-run
connectedness over long-run connectedness is observed in the net directional connectedness.
Network plots show the dominance of corn, soybean, and wheat as short-run and long-run
shock transmitters at lower and upper quantile points, while at the normal market, the
connectedness between agricultural commodities and oil, and among agricultural commodities
is disentangled in the case of short-run. The results have significant ramifications for
diversifying a portfolio and investment trading strategies. Findings in this paper, therefore,
agree with Lu et al. (219), Naeem et al. (2022), Montasser, Belhoula, and Charfeddine (2023),
and Furuoka et al. (2023) on the dominance of corn, soybean and wheat as possible net
transmitters in the oil or energy to agricultural commodity connectedness, and the weak role of
energy and oil in the network.
Future studies can extend the existing work by examining the effect of various
uncertainties and their prominent role in the association between agriculture commodities and
crude oil over different investment horizons.

References

18
Adam, P., Rianse, U., Harafah, Ĺ. M., Cahyono, E. and Rafiy, M. (2016). A model of the
dynamics of the effect of world crude oil price and world rice price on Indonesia’s
inflation rate. Agris On-Line Papers in Economics and Informatics, 8, 3–12.
Ahmadi, M., Behmiri, N. B., and Manera, M. (2016). How is volatility in commodity markets
linked to oil price shocks? Energy Economics, 59: 11–23.
Antonakakis, N. and Gabauer, D. (2017). Refined measures of dynamic connectedness based
on TVP-VAR. MPRA Paper 78282, University Library of Munich, Germany.
Balcilar, M., Gabauer, D. and Umar, Z. (2021). Crude Oil futures contracts and commodity
markets: New evidence from a TVP-VAR extended joint connectedness approach.
Resources Policy, 73, 102219.
Barunik, J., & Krehlik, T. (2018). Measuring the frequency dynamics of financial
connectedness and systemic risk. Journal of Financial Econometrics, 16(2), 271-296.
Bellmann, C. and Hepburn, J. (2017). The Decline of Commodity Prices and Global
Agricultural Trade Negotiations: A Game Changer? International Development Policy,
8.1.
Boroumand, R. H. and Porcher, T. (2023). Volatility contagion and connectedness between
WTI and commodity markets. Finance Research Letters, 58(A), 103959.
Coskun, Y., Akinsomi, O., Gil-Alana, L. A. and Yaya, O. S. (2023). Stock market responses
to COVID-19: The behaviors of mean reversion, dependence and persistence. Heliyon,
9, e15084.
Chen, S. S. (2010). Do higher oil prices push the stock market into bear territory? Energy
Economics, 32: 490–495.
Chatziantoniou, I., Gabauer, D., & Stenfors, A. (2021). Interest rate swaps and the transmission
mechanism of monetary policy: A quantile connectedness approach. Economics
Letters, 204, 109891.
Chatziantoniou, I., Abakah, E. J., Gabauer, D., & Tiwari, A. K. (2021). Quantile time-
frequency price connectedness between green bond, green equity, sustainable
investments and clean energy markets: Implications for eco-friendly investors.
Available at SSRN 3970746.
Cui, J., and Maghyereh, A. (2023). Higher-order moment risk connectedness and optimal
investment strategies between international oil and commodity futures markets:
Insights from the COVID-19 pandemic and Russia-Ukraine conflict. International
Review of Financial Analysis, 86, 102520.
Dahl, R. E., Oglend, A. and Yahya, M. (2020). Dynamics of volatility spillover in commodity
markets: Linking crude oil to agriculture. Journal of Commodity Markets, 20, 100111.
Diebold, F. X., and Yılmaz, K. (2012). Better to give than to receive: Predictive directional
measurement of volatility spillovers. International Journal of Forecasting, 28(1), 57–
66.
Diebold, F.X. and Yılmaz, K. (2014). On the network topology of variance decompositions:
Measuring the connectedness of financial firms. Journal of Econometrics, 182(1), 119–
134.
Furuoka, F., Yaya, O. S., Ling, P. K., Al-Faryan, M. A. S. nd Islam, M. N. (2023). Transmission
of risks between energy and agricultural commodities: Frequency time-varying VAR,
asymmetry and portfolio management. Resources Policy, 81, 103339.
Gong, X, Jin, Y., and Liu, T. (2023). Analyzing pure contagion between crude oil and
agricultural futures markets. Energy 269, 126757.
Guhathakurta, K., Dash, S. R. and Maitra, D. (2020). Period specific volatility spillover based
19
 connectedness between oil and other commodity prices and their portfolio implications,
Energy Economics, 85, 104566.
Hau, L., Zhu, H., Huang, R. and Ma, X. (2020). Heterogeneous dependence between crude oil
price volatility and China’s agriculture commodity futures: Evidence from quantile-on-
quantile regression. Energy, 213, 118781.
Hung, N. T. (2021). Oil prices and agricultural commodity markets: Evidence from pre and
during COVID-19 outbreak. Resources Policy, 73, 102236.
IEA (2022). Imports/Exports, https://www.iea.org/data-and-statistics. Accessed on
12/01/2024.
Ji, Q., Bouri, E., Roubaud, D., and Shahzad, S. J. H. (2018). Risk spillover between energy
and agricultural commodity markets: A Dependence-switching CoVaR-Copula
Model. Energy Economics, 75, 14–27.
Kang, S. H., Tiwari, A. K., Albulescu, C. T., and Yoon, S. M. (2019). Exploring the time-
frequency connectedness and network among crude oil and agriculture commodities.
Energy Economics, 84, 104543.
Koirala, K. H., Mishra, A. K., D’Antoni, J. M., and Mehlhorn, J. E. (2015). Energy prices and
agricultural commodity prices: Testing correlation using copulas method. Energy, 81:
430–436.
Koop, G., Pesaran, M. H., and Potter, S. M. (1996). Impulse response analysis in nonlinear
multivariate models. Journal of Econometrics, 74(1), 119-147.
Londono, J.M. (2019). Bad bad contagion. Journal of Banking and Finance, 108, 105652.
Lu, Y., Yang L., and Liu L. (2019). Volatility spillovers between crude oil and agricultural
commodity markets since the financial crisis. Sustainability, 11(2), 396.
Maitra, D., Guhathakurta, K., and Kang, S. H. (2021). The good, the bad and the ugly relation
between oil and commodities: An analysis of asymmetric volatility connectedness and
portfolio implications. Energy Economics, 94, 105061,
Montasser, G. E., Belhoula, M. M., Charfeddine, L. (2023). Co-explosivity versus leading
effects: Evidence from crude oil and agricultural commodities. Resources Policy, 81,
103331.
Naeem, M. A., Karim, S., Hasan, M., Lucey, B. M. and Kang, S. H. (2022). Nexus between oil
shocks and agriculture commodities: Evidence from time and frequency domain,
Energy Economics, 112, 106148.
Oil World (2022). https://www.oilworld.biz/. Accessed on 12/01/2024.
Pal, D., and Mitra, S. K. (2020). Time-frequency dynamics of return spillover from crude oil
to agricultural commodities. Applied Economics, 52(49), 5426-5445.
Pesaran, M.H. and Shin, Y. (1998). Generalized Impulse Response Analysis in Linear
Multivariate Models. Economics Letters, 58, 17-29.
Rafiq, S., Salim, R., and Bloch, H. (2009). Impact of crude oil price volatility on economic
activities: An empirical investigation in the Thai economy. Resources Policy, 34(3):
121–132.
Reboredo, J. C. (2012). Do food and oil prices co-Move? Energy Policy 49, 456–67.
Rezitis, A. N. (2015). The relationship between agricultural commodity prices, crude oil
prices and US dollar exchange rates: A panel VAR approach and causality analysis.
International Review of Applied Economics 29(3), 403–34.
Serletis, A. and Xu, L. (2019). The ethanol mandate and crude oil and biofuel agricultural
commodity price dynamics. Journal of Commodity Markets, 15, 100068.
20
Shahzad, S. J.H., Hernandez, J., Al-Yahyaee, K., and Jammazi. R. (2018). Asymmetric risk
spillovers between oil and agricultural commodities. Energy Policy 118, 182–98.
Stiassny, A., 1996. A spectral decomposition for structural VAR models. Empirical
Economics, 21(4), 535–555.
Sun, Y., Mirza, N., Qadeer, A., and Hsueh, H.-P. (2021). Connectedness between oil and
agricultural commodity prices during tranquil and volatile periods. Is crude oil a victim
indeed? Resources Policy, 72, 102131.
Tiwari, A. K., Abakah, E. J. A., Adewuyi, A. O. and Lee, C.-C. (2022). Quantile risk spillovers
between energy and agricultural commodity markets: Evidence from pre and during
COVID-19 outbreak. Energy Economics, 113, 106235.
Tiwari, A. K., Khalfaoui, R., Solarin, S. A. and Shahbaz, M. (2018). Analyzing the time-
frequency lead–lag relationship between oil and agricultural commodities. Energy
Economics, 76, 470-494.
Umar, Z., Jareño, F. and Escribano, A. (2021). Agricultural commodity markets and oil prices:
An analysis of the dynamic return and volatility connectedness. Resources Policy, 73,
102147.
Vacha, L., Janda, K., Kristoufek, L., & Zilberman, D., (2013). Time-frequency dynamics of
the biofuel-fuel-food system. Energy Economics, 40: 233–241.
Wang, K.-H., Kan, J.-M., Qiu, L. and Xu, S. (2023). Climate policy uncertainty, oil price and
agricultural commodity: From quantile and time perspective. Economic Analysis and
Policy, 78, 256-272.
Wu, Y., Ren, W., Wan, J. and Liu, X. (2023). Time-frequency volatility connectedness between
fossil energy and agricultural commodities: Comparing the COVID-19 pandemic with
the Russia-Ukraine conflict. Finance Research Letters, 55(A), 103866.
Yaya, O. S., Abu, N. and Ogundunmade, T. P. (2021). Economic Policy Uncertainty in G7
countries: Evidence of Long-range Dependence and Cointegration. Economic Change
and Restructuring, 54(2), 541-556.
Yaya, O. S., Olayinka, H. A., Ogbonna, E. A., Al-Faryan, M. A. S. and Vo, X. V. (2024).
Dynamic connectedness of Economic Policy Uncertainty in G7 countries, and the
influence of the US and UK: Quantile VAR analysis. Economic Change and
Restructuring, 57(2), 1-27.
Yoon, S.-M. (2022). On the interdependence between biofuel, fossil fuel and agricultural food
prices: Evidence from quantile tests. Renewable Energy, 199, 536-545.
Zhang, C., and Qu, X. (2015). The effect of global oil price shocks on China's agricultural
commodities. Energy Economics, 51, 354-364.
Živkov, D., Manić, S. and Đurašković, J. (2020). Short and long-term volatility transmission
from oil to agricultural commodities – The robust quantile regression approach. Borsa
Istanbul Review, 20, S11-S25.

21
 Brent WTI Wheat Corn
160 150 2,000 800

1,600
120 100 600
1,200
80 50 400
800
40 0 200
400

0 -50 0 0
90 95 00 05 10 15 20 90 95 00 05 10 15 20 90 95 00 05 10 15 20 90 95 00 05 10 15 20

Soybeans Cotton Sugar Coffee

8,000 1,600 400 800

6,000 1,200 300 600

4,000 800 200 400

2,000 400 100 200

0 0 0 0
90 95 00 05 10 15 20 90 95 00 05 10 15 20 90 95 00 05 10 15 20 90 95 00 05 10 15 20

Cocoa Live cattle Lean hogs

100 6,000 2,000

80 5,000 1,600

60 4,000 1,200

40 3,000 800

20 2,000 400

0 1,000 0
90 95 00 05 10 15 20 90 95 00 05 10 15 20 90 95 00 05 10 15 20

Figure 1: Time plots of WTI, Brent, and agricultural commodities prices

22
Table 1: Preliminary analyses and tests

Descriptive Brent WTI Wheat Corn Soybeans Cotton Sugar Coffee Cocoa Live. cattle Lean. hogs
Mean 0.00015 0.00021 -0.00021 -0.00013 0.00022 0.00001 0.00016 -0.00016 -0.00006 0.00011 -0.00021
Variance 0.00062 0.00067 0.00029 0.00023 0.00019 0.00023 0.00037 0.00047 0.00031 0.00008 0.00025
Skewness -1.830*** -0.004 0.083*** -0.005 -0.168*** -0.059** -0.202*** 0.269*** -0.013 -0.131*** -0.072***
Ex. Kurtosis 68.527*** 29.466*** 3.441*** 2.793*** 2.707*** 1.524*** 2.965*** 6.705*** 2.539*** 2.365*** 1.642***
JB 1872160.6*** 345154.6*** 4719.3*** 3100.7*** 2957.3*** 929.4*** 3559.2*** 17987.6*** 2563.5*** 2251.4*** 1080.2***
ERS -31.963*** -33.319*** -43.698*** -12.595*** -12.582*** -9.614*** -10.112*** -36.732*** -43.201*** -10.833*** -19.173***
Q (20) 72.62*** 68.622*** 13.577 37.632*** 20.710** 32.170*** 21.085** 29.417*** 14.099 58.240*** 36.790***
Q2(20) 1788.34*** 3066.21*** 1708.72*** 2227.40*** 2036.44*** 2144.83*** 723.15*** 1120.52*** 323.84*** 4081.60*** 3594.30***
Pearson Correlations Brent WTI Wheat Corn Soybeans Cotton Sugar Coffee Cocoa Live. cattle Lean.hogs
Brent 1.000*** 0.587*** 0.081*** 0.107*** 0.108*** 0.132*** 0.105*** 0.069*** 0.090*** 0.053*** 0.019
WTI 0.587*** 1.000*** 0.128*** 0.155*** 0.163*** 0.142*** 0.128*** 0.086*** 0.082*** 0.087*** 0.069***
Wheat 0.081*** 0.128*** 1.000*** 0.610*** 0.457*** 0.193*** 0.164*** 0.120*** 0.080*** 0.070*** 0.080***
Corn 0.107*** 0.155*** 0.610*** 1.000*** 0.656*** 0.213*** 0.176*** 0.119*** 0.079*** 0.082*** 0.081***
Soybeans 0.108*** 0.163*** 0.457*** 0.656*** 1.000*** 0.240*** 0.182*** 0.135*** 0.103*** 0.092*** 0.092***
Cotton 0.132*** 0.142*** 0.193*** 0.213*** 0.240*** 1.000*** 0.145*** 0.099*** 0.087*** 0.069*** 0.068***
Sugar 0.105*** 0.128*** 0.164*** 0.176*** 0.182*** 0.145*** 1.000*** 0.148*** 0.121*** 0.061*** 0.058***
Coffee 0.069*** 0.086*** 0.120*** 0.119*** 0.135*** 0.099*** 0.148*** 1.000*** 0.148*** 0.052*** 0.031***
Cocoa 0.090*** 0.082*** 0.080*** 0.079*** 0.103*** 0.087*** 0.121*** 0.148*** 1.000*** 0.047*** 0.029***
Live. cattle 0.053*** 0.087*** 0.070*** 0.082*** 0.092*** 0.069*** 0.061*** 0.052*** 0.047*** 1.000*** 0.305***
Lean.hogs 0.019 0.069*** 0.080*** 0.081*** 0.092*** 0.068*** 0.058*** 0.031*** 0.029*** 0.305*** 1.000***
Persistence tests Brent WTI Wheat Corn Soybeans Cotton Sugar Coffee Cocoa Live. cattle Lean.hogs
Simple R/S Hurst. 0.533 0.520 0.524 0.523 0.497 0.552 0.522 0.509 0.510 0.542 0.544
Corrected Hurst. 0.564 0.551 0.547 0.556 0.528 0.585 0.538 0.540 0.511 0.556 0.561
Empirical Hurst. 0.565 0.560 0.540 0.558 0.521 0.598 0.540 0.536 0.486 0.538 0.554
Corrected empirical Hurst. 0.538 0.532 0.513 0.530 0.494 0.569 0.512 0.508 0.458 0.510 0.526
Theoretical Hurst. 0.527 0.527 0.527 0.527 0.527 0.527 0.527 0.527 0.527 0.527 0.527
Note, ***, ** and * indicate significance at 1, 5 and 10% level, respectively. Note, that persistence tests are required here to complement Pearson correlation in further testing
for interconnectedness in returns through cross-sectional dependence of returns. Stationarity of returns based on ERS or ADF unit root tests will not bring out this salient
property of persistence in returns which is used here.

23
Table 2: Average connectedness at the lower quantile, 𝝉 = 𝟎. 𝟎𝟓

Soy Live. Lean.

Brent WTI Wheat Corn Cotton Sugar Coffee Cocoa FROM
beans cattle hogs
Brent 13.97, 11.68, 8.2, 8.24, 8.44, 8.42, 8.21, 8.32, 8.31, 8.12, 8.1, 86.03,
10.27, 3.7 8.37, 3.31 5.95, 2.24 5.92, 2.31 6.13, 2.31 6.18, 2.24 5.92, 2.29 6.09, 2.23 5.98, 2.34 5.95, 2.17 5.92, 2.18 62.41, 23.62

WTI 11.08, 14.49, 8.21, 8.28, 8.42, 8.42, 8.35, 8.29, 8.2, 8.09, 8.18, 85.51,
8.33, 2.75 11.07, 3.41 6.22, 1.99 6.17, 2.11 6.32, 2.1 6.34, 2.08 6.22, 2.13 6.24, 2.05 6.06, 2.14 6.11, 1.98 6.13, 2.05 64.13, 21.38

Wheat 7.9, 8.33, 14.22, 10.73, 9.82, 8.41, 8.28, 8.22, 8.01, 8.0, 8.08, 85.78,
6.29, 1.61 6.69, 1.64 11.47, 2.75 8.6, 2.13 8.0, 1.82 6.76, 1.65 6.6, 1.68 6.6, 1.62 6.35, 1.66 6.45, 1.55 6.5, 1.58 68.84, 16.94

Corn 7.87, 8.3, 10.53, 13.9, 10.93, 8.34, 8.19, 8.16, 7.88, 7.88, 8.03, 86.1,
5.94, 1.92 6.34, 1.96 8.03, 2.5 10.6, 3.3 8.43, 2.5 6.30, 2.04 6.15, 2.04 6.13, 2.04 5.89, 1.99 5.99, 1.89 6.04, 2.0 65.23, 20.88
Soy 8.03, 8.42, 9.68, 10.94, 14.02, 8.48, 8.18, 8.21, 8.02, 7.95, 8.08, 85.98,
beans 6.23, 1.8 6.6, 1.81 7.64, 2.04 8.48, 2.46 11.07, 2.95 6.63, 1.85 6.39, 1.8 6.4, 1.81 6.2, 1.82 6.21, 1.74 6.3, 1.78 67.08, 18.9

Cotton 8.41, 8.64, 8.67, 8.77, 8.91, 14.74, 8.47, 8.53, 8.32, 8.16, 8.37, 85.26,
6.4, 2.01 6.63, 2.01 6.67, 2.0 6.64, 2.13 6.86, 2.05 11.39, 3.35 6.47, 2.0 6.53, 2.0 6.28, 2.04 6.32, 1.84 6.42, 1.95 65.23, 20.03

Sugar 8.22, 8.73, 8.65, 8.67, 8.75, 8.72, 14.48, 8.69, 8.5, 8.19, 8.4, 85.52,
6.22, 2.0 6.67, 2.06 6.57, 2.09 6.45, 2.22 6.64, 2.11 6.59, 2.13 11.08, 3.39 6.6, 2.09 6.32, 2.18 6.2, 1.99 6.3, 2.1 64.56, 20.96

Coffee 8.33, 8.62, 8.51, 8.46, 8.57, 8.5, 8.75, 14.8, 8.83, 8.27, 8.37, 85.2,
6.42, 1.91 6.72, 1.9 6.66, 1.85 6.54, 1.92 6.74, 1.83 6.67, 1.83 6.8, 1.95 11.74, 3.06 6.78, 2.05 6.49, 1.78 6.49, 1.89 66.31, 18.89

Cocoa 8.44, 8.65, 8.4, 8.35, 8.56, 8.44, 8.61, 8.9, 14.97, 8.28, 8.4, 85.03,
6.03, 2.4 6.23, 2.42 6.06, 2.34 5.96, 2.39 6.22, 2.34 6.15, 2.3 6.14, 2.47 6.46, 2.43 10.95, 4.02 6.01, 2.27 6.07, 2.33 61.33, 23.7
Live. 8.26, 8.57, 8.35, 8.31, 8.44, 8.37, 8.3, 8.4, 8.36, 14.83, 9.81, 85.17,
cattle 6.24, 2.02 6.55, 2.03 6.34, 2.01 6.24, 2.07 6.43, 2.0 6.42, 1.95 6.26, 2.04 6.48, 1.92 6.27, 2.09 11.43, 3.4 7.48, 2.34 64.71, 20.46
Lean. 8.17, 8.59, 8.36, 8.42, 8.5, 8.45, 8.39, 8.33, 8.35, 9.68, 14.76, 85.24,
hogs 6.08, 2.09 6.43, 2.15 6.21, 2.14 6.21, 2.21 6.35, 2.15 6.32, 2.13 6.24, 2.15 6.28, 2.05 6.13, 2.22 7.28, 2.41 11.12, 3.63 63.54, 21.71

TO 84.7, 88.53, 87.56, 89.16, 89.32, 84.55, 83.73, 84.05, 82.78, 82.63, 83.82, 940.83,
64.19, 20.5 67.24, 21.29 66.35, 21.2 67.21, 21.94 68.11, 21.21 64.34, 20.21 63.19, 20.54 63.81, 20.24 62.26, 20.52 63.01, 19.62 63.64, 20.18 713.4, 227.5
Inc. 98.67, 103.02, 101.77, 103.06, 103.34, 99.29, 98.21, 98.85, 97.75, 97.46, 98.58,
Own 74.46, 24.2 78.32, 24.7 77.82, 23.95 77.81, 25.24 79.18, 24.16 75.73, 23.56 74.28, 23.94 75.55, 23.3 73.21, 24.55 74.44, 23.02 74.76, 23.82 TCI

Net -1.33, 3.02, 1.77, 3.06, 3.34, -0.71, -1.79, -1.15, -2.25, -2.54, -1.42, 85.53,
1.79, -3.12 3.11, -0.09 -2.49, 4.26 1.99, 1.07 1.03, 2.31 -0.88, 0.18 -1.37, -0.42 -2.5, 1.35 0.93, -3.18 -1.7, -0.84 0.10, -1.52 64.85, 20.68
Note, in each cell, the upper value is the total connectedness rendered by the time connectedness measure, while the two lower values in the cell are the short- and long-term
frequency connectedness measures.

24
Table 3: Average connectedness at the upper quantile, 𝝉 = 𝟎. 𝟗𝟓

Live. Lean.
Brent WTI Wheat Corn Soybeans Cotton Sugar Coffee Cocoa FROM
cattle hogs
Brent 14.74, 11.65, 8.12, 8.2, 8.15, 8.43, 8.22, 7.98, 8.19, 8.27, 8.04, 85.26,
11.77, 2.97 9.06, 2.59 6.44, 1.69 6.49, 1.71 6.46, 1.69 6.69, 1.75 6.53, 1.7 6.35, 1.63 6.52, 1.67 6.55, 1.72 6.36, 1.68 67.46, 17.8
WTI 11.31, 14.99, 8.21, 8.29, 8.26, 8.36, 8.24, 8.01, 8.07, 8.23, 8.03, 85.01,
9.24, 2.07 12.18, 2.81 6.63, 1.58 6.7, 1.59 6.69, 1.57 6.79, 1.57 6.7, 1.53 6.52, 1.49 6.56, 1.51 6.65, 1.58 6.53, 1.51 69.01, 15.99
Wheat 7.89, 8.11, 14.35, 10.8, 9.81, 8.51, 8.27, 8.12, 7.95, 8.14, 8.06, 85.65,
6.06, 1.83 6.24, 1.87 10.98, 3.37 8.29, 2.51 7.51, 2.3 6.47, 2.04 6.38, 1.89 6.2, 1.91 6.1, 1.84 6.3, 1.84 6.22, 1.85 65.77, 19.89
Corn 7.83, 8.02, 10.56, 14.16, 11.0, 8.51, 8.16, 7.97, 7.8, 8.03, 7.97, 85.84,
6.35, 1.48 6.53, 1.49 8.53, 2.03 11.38, 2.78 8.85, 2.15 6.81, 1.71 6.61, 1.55 6.44, 1.53 6.32, 1.47 6.54, 1.49 6.48, 1.49 69.45, 16.4
Soy
7.79, 8.12, 9.78, 11.15, 14.3, 8.58, 8.29, 8.02, 7.87, 8.11, 7.99, 85.7,
beans
6.26, 1.53 6.54, 1.58 7.91, 1.88 9.02, 2.13 11.51, 2.79 6.88, 1.7 6.67, 1.62 6.45, 1.57 6.32, 1.55 6.57, 1.54 6.47, 1.51 69.09., 16.61
Cotton 8.34, 8.41, 8.7, 8.85, 8.86, 14.96, 8.56, 8.34, 8.18, 8.47, 8.34, 85.04,
6.72, 1.61 6.78, 1.63 6.94, 1.76 7.1, 1.75 7.13, 1.73 11.95, 3.02 6.87, 1.69 6.67, 1.66 6.59, 1.59 6.85, 1.63 6.75, 1.59 68.39, 16.65
Sugar 8.21, 8.49, 8.64, 8.63, 8.65, 8.65, 14.91, 8.66, 8.43, 8.46, 8.27, 85.09,
6.44, 1.77 6.63, 1.86 6.77, 1.87 6.74, 1.89 6.79, 1.87 6.78, 1.88 11.79, 3.13 6.77, 1.89 6.68, 1.75 6.62, 1.84 6.49 1.78 66.7, 18.39
Coffee 8.1, 8.35, 8.54, 8.54, 8.65, 8.55, 8.79, 15.05, 8.66, 8.48, 8.32, 84.95,
6.38, 1.72 6.61, 1.74 6.62, 1.92 6.7, 1.83 6.75, 1.9 6.7, 1.85 6.9, 1.89 11.89, 3.16 6.86, 1.79 6.64, 1.83 6.51, 1.81 66.67, 18.28
Cocoa 8.3, 8.36, 8.46, 8.38, 8.47, 8.53, 8.54, 8.77, 15.23, 8.59, 8.36, 84.77,
6.62, 1.68 6.68, 1.68 6.71, 1.75 6.7, 1.68 6.73, 1.74 6.76, 1.77 6.82, 1.72 6.97, 1.8 12.18, 3.05 6.84, 1.76 6.67, 1.69 67.5, 17.27
Live.
8.15, 8.31, 8.38, 8.36, 8.42, 8.48, 8.28, 8.26, 8.27, 15.32, 9.76, 84.68,
cattle
6.69, 1.46 6.84, 1.47 6.82, 1.56 6.86, 1.5 6.9,1.52 6.96, 1.53 6.76, 1.52 6.75, 1.52 6.75, 1.52 12.5, 2.82 8.05, 1.71 69.39, 15.29
Lean.
8.03, 8.29, 8.41, 8.37, 8.42, 8.61, 8.34, 8.25, 8.24, 9.96, 15.07, 84.93,
hogs
6.23, 1.8 6.47, 1.82 6.51, 1.9 6.54, 1.84 6.53, 1.89 6.72, 1.89 6.45, 1.88 6.44, 1.81 6.42, 1.82 7.77, 2.2 11.8, 3.27 66.08, 18.85
TO 10.4,
83.95, 86.13, 87.8, 89.57, 88.69, 85.22, 83.69, 82.37, 81.64, 83.14, 936.92,
67.0, 16.95 68.3817.74 69.87, 17.93 71.14, 18.43 70.33, 18.36 67.56, 17.66 66.69, 16.99 65.56, 16.81 65.14, 16.51 7.29, 3.11 66.53, 16.62 745.51, 191.42
Inc.
98.69, 101.12, 102.14, 103.73, 102.98, 100.18, 98.6, 97.42, 96.87, 100.05, 98.22,
Own
78.77, 19.92 80.56, 20.55 80.85, 21.3 82.51, 21.21 81.84, 21.15 79.5, 20.67 78.48, 20.12 77.45, 19.97 77.32, 19.56 79.81, 20.24 78.33, 19.89 TCI
Net
-1.31, 1.12, 2.14, 3.73, 2.98, 0.18, -1.4, -2.58, -3.13, 0.05, -1.78, 85.17,
-0.46, -0.85 -0.63, 1.75 4.1, -1.96 1.69, 2.04 1.23, 1.75 -0.83, 1.01 -0.01, -1.4 -1.11, -1.48 -2.36, -0.76 -2.08, 2.13 0.45, -2.23 67.77, 17.4
Note, in each cell, the upper value is the total connectedness rendered by the time connectedness measure, while the two lower values in the cell are the short- and long-term
frequency connectedness measures.

25
Table 4: Average connectedness at the middle quantile, 𝝉 = 𝟎. 𝟓

Live. Lean.
Brent WTI Wheat Corn Soybeans Cotton Sugar Coffee Cocoa FROM
cattle hogs
Brent 58.07, 28.16, 1.49, 1.63, 2.07, 1.8, 1.77, 1.43, 1.49, 1.26, 0.84, 41.93,
46.2, 11.88 20.44, 7.72 1.16, 0.34 1.26, 0.36 1.54, 0.53 1.45, 0.35 1.38, 0.39 1.10, 0.32 1.19, 0.30 1.01, 0.25 0.68, 0.16 31.21, 10.71

WTI 23.42, 61.45, 1.61, 1.93, 2.38, 1.99, 2.01, 1.69, 1.42, 1.19, 0.92, 38.55,
19.05, 4.37 50.21, 11.24 1.31, 0.29 1.58, 0.35 1.89, 0.48 1.61, 0.38 1.62, 0.39 1.34, 0.35 1.16, 0.26 0.99, 0.2 0.77, 0.15 31.33, 7.22

Wheat 1.22, 1.8, 57.09, 19.41, 11.26, 2.19, 1.9, 1.58, 1.14, 1.32, 1.09, 42.91,
0.99, 0.23 1.53, 0.27 46.52, 10.58 15.84, 3.57 9.39, 1.86 1.8, 0.39 1.57, 0.34 1.29, 0.3 0.96, 0.18 1.06, 0.26 0.87, 0.22 35.29, 7.62

Corn 1.28, 1.84, 17.3, 50.01, 21.17, 2.31, 1.79, 1.44, 0.95, 1.03, 0.89, 49.99,
1.07, 0.21 1.55, 0.29 13.93, 3.38 40.29, 9.71 17.23, 3.93 1.87, 0.44 1.39, 0.4 1.15, 0.29 0.77, 0.18 0.84, 0.19 0.71, 0.17 40.51, 9.48
Soy 1.6, 2.36, 10.54, 22.41, 53.07, 2.98, 2.09, 1.84, 1.06, 1.02, 1.03, 46.93,
beans 1.37, 0.23 2.02, 0.33 8.8, 1.74 18.68, 3.74 43.87, 9.2 2.48, 0.5 1.73, 0.36 1.49, 0.35 0.87, 0.19 0.82, 0.2 0.84 0.18 39.11, 7.82

Cotton 2.29, 2.59, 2.8, 3.26, 3.9, 77.01, 2.31, 2.14, 1.33, 1.18, 1.18, 22.99,
1.87, 0.43 2.08, 0.51 2.29, 0.51 2.59, 0.68 3.14, 0.76 63.28, 13.74 1.88, 0.43 1.73, 0.41 1.08, 0.25 1.01, 0.17 0.97 0.21 18.63, 4.36

Sugar 2.05, 2.69, 2.33, 2.54, 3.0, 2.29, 77.64, 3.3, 1.9, 1.13, 1.14, 22.36,
1.71, 0.34 2.25, 0.43 1.88, 0.45 1.95, 0.59 2.34, 0.66 1.86, 0.43 64.23, 13.41 2.7, 0.6 1.58, 0.32 0.94, 0.18 0.97, 0.17 18.17, 4.19

Coffee 1.49, 2.04, 1.88, 1.96, 2.57, 2.21, 3.32, 79.44, 2.69, 1.38, 1.01, 20.56,
1.22, 0.27 1.65, 0.39 1.49, 0.38 1.57, 0.39 2.01, 0.56 1.8, 0.41 2.75, 0.58 65.77, 13.67 2.26, 0.42 1.17, 0.21 0.85, 0.16 16.77, 3.79

Cocoa 1.9, 1.99, 1.4, 1.42, 1.82, 1.54, 2.04, 2.98, 82.64, 1.25, 1.03, 17.36,
1.53, 0.37 1.59, 0.4 1.12, 0.29 1.1, 0.31 1.43, 0.39 1.2, 0.34 1.62, 0.43 2.4, 0.58 68.23, 14.42 1.02, 0.22 0.86, 0.17 13.86, 3.5
Live. 1.43, 1.62, 1.55, 1.56, 1.76, 1.3, 1.11, 1.36, 1.26, 78.99, 8.06, 21.01,
Cattle 1.19, 0.24 1.35, 0.27 1.29, 0.26 1.32, 0.23 1.45, 0.31 1.09, 0.21 0.93, 0.18 1.13, 0.23 1.04, 0.22 64.37, 14.62 6.68, 1.38 17.46, 3.54
Lean. 1.04, 1.41, 1.39, 1.52, 1.82, 1.28, 1.27, 1.06, 1.01, 8.35, 79.86, 20.14,
Hogs 0.88, 0.16 1.2, 0.21 1.15, 0.24 1.2, 0.32 1.48, 0.35 1.06, 0.21 1.02, 0.25 0.89, 0.17 0.86, 0.16 7.01, 1.34 65.44, 14.42 16.73, 3.41

TO 37.73, 46.49, 42.3, 57.64, 51.73, 19.87, 19.62, 18.82, 14.25, 19.1, 17.18, 344.72,
30.87, 6.86 35.66, 10.83 34.41, 7.88 47.1, 10.54 41.9, 9.84 16.2, 3.67 15.89, 3.73 15.21, 3.61 11.77, 2.48 15.87, 3.23 14.2, 2.98 279.08, 65.64
Inc. 95.8, 107.94, 99.39, 107.65, 104.8, 96.88, 97.26, 98.26, 96.89, 98.09, 97.03,
Own 77.07, 18.74 85.88, 22.07 80.93, 18.46 87.39, 20.26 85.77, 19.04 79.48, 17.4 80.13, 17.13 80.98, 17.27 80, 16.89 80.24. 17.85 79.64, 17.4 TCI

Net -4.2, 7.94, -0.6,1 7.65, 4.8, -3.12, -2.74, -1.74, -3.11, -1.91, -2.97, 31.34,
-0.35, -3.85 4.34, 3.61 -0.8, 0.26 6.58, 1.06 2.79, 2.01 -2.42, -0.69 -2.28, -0.47 -1.57, -0.18 -2.09, -1.02 -1.6, -0.31 -2.54, -0.43 25.37, 5.97
Note, in each cell, the upper value is the total connectedness rendered by the time connectedness measure, while the two lower values in the cell are the short- and long-term
frequency connectedness measures.

26
Figure 2: Dynamic Connectedness at the lower quantile
Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

Figure 3: Net Directional Connectedness at the lower quantile

Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

27
Figure 4: Dynamic Connectedness at the upper quantile
Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

Figure 5: Net Directional Connectedness at the upper quantile

Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

28
Figure 6: Dynamic Connectedness at the middle quantile
Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

Figure 7: Net Directional Connectedness at the middle quantile

Note, the black part represents the total connectedness rendered by time connectedness while the green and red
parts represent long- and short-term connectedness rendered by frequency connectedness.

29
 Total

Long-term
Short-term

Figure 8: Network plots at lower quantile

Note, blue-coloured nodes are net shock transmitters while orange-coloured nodes are net shock receivers.

30
 Total

Long-term

Short-term

Figure 9: Network plots at the upper quantile

Note, blue-coloured nodes are net shock transmitters while orange-coloured nodes are net shock receivers.

31
Total

Long-term

Short-term

Figure 10: Network plots at middle quantile

Note, blue-coloured nodes are net shock transmitters while orange-coloured nodes are net shock receivers.

32