Estimating the connectedness of commodity futures using a network approach

• Author: Sifang Ding,Libing Fang,Binqing Xiao,Honghai Yu
• Date: 01 April 2020
• DOI: http://doi.org/10.1002/fut.22086
• Published date: 01 April 2020

J Futures Markets. 2020;40:598–616.wileyonlinelibrary.com/journal/fut598

|

© 2019 Wiley Periodicals, Inc.

Received: 2 April 2019

|

Accepted: 2 December 2019

DOI: 10.1002/fut.22086

RESEARCH ARTICLE

Estimating the connectedness of commodity futures using

a network approach

Binqing Xiao

|

Honghai Yu

|

Libing Fang

|

Sifang Ding

School of Management and Engineering,

Nanjing University, Nanjing, Jiangsu,

China

Correspondence

Libing Fang, School of Management and

Engineering, Nanjing University, No. 22

Hankou Road, Gulou District, Jiangsu

Province, 210093 Nanjing, China.

Email: lbfang@nju.edu.cn

Funding information

National Science Foundation of China,

Grant/Award Numbers: U1811462,

71720107001, 71671083, 71771117,

71972100

Abstract

Using a network approach of variance decompositions, we measure the

connectedness of 18 commodity futures and characterize both static and

dynamic connectedness. Our results show that metal futures are net

transmitters of shocks to other futures, and agricultural futures are vulnerable

to shocks from the others. Furthermore, almost two‐thirds of the volatility

uncertainty for commodity futures are due to the connectedness of shocks

across the futures market. Dynamically, we find connectedness always increases

in times of turmoil. An analysis of connectedness networks suggests that

investors could be forewarned that the connectedness of various classes of

futures could threaten their portfolios.

KEYWORDS

commodity futures, connectedness, variance decompositions

1

|

INTRODUCTION

The commodities futures market plays an important role in the global economy. Due to low correlations with

traditional financial assets and positive comovements of commodity prices with inflation, commodity futures, as a

profitable hedging and speculation asset, can help investors optimize asset allocation and diversify portfolio risks

(Chong & Miffre, 2010; Kang, McIver, & Yoon, 2017; Silvennoinen & Thorp, 2013). There are three main types of

commodity futures—metal futures, energy futures, and agriculture futures. The financialization of commodity markets

is likely to increase the degree of integration across different commodity markets, which is a significant development

for investors and policymakers (Kang et al., 2017; Mensi, Hammoudeh, Nguyen, & Yoon, 2014; Tang & Xiong, 2012).

Connectedness, as an indicator of the correlations between market factors, can be used to quantitatively measure

systemic risk. The global financial crisis of 2007–2009 has promoted increased interest in systemic risk,

1

which is now

broadly applied to study shocks to other parts of the financial system (Billio, Getmansky, Lo, & Pelizzon, 2012; Kang

et al., 2017). The connectedness is crucial to modern financial risk measurement and management (Demirer, Diebold,

Liu, & Yilmaz, 2018; Diebold, Liu, & Yilmaz, 2017; Diebold & Yilmaz, 2014). Some articles study connectedness in asset

prices using minimum spanning trees (MSTs) and other network representations in stock markets and currency

markets considering idiosyncratic versus systematic risk of connectedness (Bonanno, Caldarelli, Lillo, & Mantegna,

2003; De Carvalho & Gupta, 2017; Mantegna, 1999; Mizuno, Takayasu, & Takayasu, 2006). Many aspects of

connectedness can be measured within a system, from pairwise to system‐wide (Diebold & Yilmaz, 2014). Therefore,

1

In this paper, we mainly focus on systemic risk it. As for the systematic risk, also known as market risk or nondistributable risk, it refers to the risk caused byfactors that affect all companies in the

market. Systematic risks are caused by external factors of the company, such as wars, regime changes, natural disasters, economic cycles, inflation, energy crises, macroeconomic policy adjustments,

and so forth.

the core of connectedness is the interaction of factors at the same level within a system rather than other driving factors,

such as macroeconomic variables. However, although volatility spillovers between stock and commodity markets have

been studied extensively, there is little empirical research on the interrelationships among different types of commodity

futures; therefore, this new area of research may provide interesting insights.

As stated in IOSCO

2

(2011), systemic risk refers to the risk of failure or failure of the entire system, which can be

quantitatively described by market index correlations, such as connectedness. Connectedness networks appear central

to modern risk measurement and management. Moreover, connectedness networks are central to understanding

underlying fundamental macroeconomic risks; they feature prominently in key aspects of market risk (return

connectedness and portfolio concentration), credit risk (default connectedness), counter‐party, and, not least, systemic

risk (system‐wide connectedness).

Commodity futures markets, as the large part of global financial market, provide major risk hedging vehicles for real

economy activities. Connectedness of commodity prices and links between commodity price connectedness are central to

risk measurement and management. In this paper, we use the network approach developed in Diebold and Yilmaz (2009,

2012, 2014) to measure the connectedness of 18 key commodity futures markets. This method captures the direction of

shock transmissions based on forecast error variance decompositions (FEVDs), which are estimated by vector

autoregressions (VARs). There are some important advantages to the network approach. First, this approach builds a

directed volatility spillover network and provides important information for analyzing dynamic directional spillovers of

futures marketsinto each other (Shahzad, Kayani,Raza, Shah, & Al‐Yahyaee, 2018). Second, it allows measurement of the

returns or volatility spilloversacross a relatively large varietyof assets and markets (Batten,Ciner, & Lucey, 2015; Zhang &

Wang, 2014). Third, Diebold and Yilmaz (2014) stress that estimates of variance decomposition are not impacted by the

ordering of the variables due to invariant FEVDs, which are independent of the Cholesky factor identification of VAR.

Due to these significant advantages, we make use of the network approach to study the systemic risk within commodity

futures markets and seek to find the connectedness evolution for 18 important commodity futures.

Both static (full‐sample) and dynamic (rolling‐sample) connectedness have unique characteristics. Throughout the

full sample, the share of volatility shocks received from other futures markets in the total variance of the forecast error

for each futures market is called “from”connectedness. In other words, the “from”connectedness means the shock

from others to each of the futures markets. Similarly, the “to”connectedness measures the shock to other futures

markets for each of the futures markets. The difference between the “to”connectedness and the “from”connectedness

yields the net total directional connectedness of each futures market to others (Diebold & Yilmaz, 2009, 2014;

Maghyereh et al., 2016). A static analysis shows that, in general, the net total connectedness of the lead futures is 34.6%,

and zinc futures have a connectedness of 30.2%, indicating that these metal futures are net transmitters of shocks to

other futures. In other words, the regulatory authorities need to formulate relevant rules to avoid shock caused by metal

futures to the connectedness network. However, the negative net total connectedness of the energy futures and

agricultural futures shows that they are relatively vulnerable to shocks from other futures markets. We find, based on

dynamic analysis, that the “from”connectedness of each futures market fluctuates smoothly over time, but the “to”

connectedness of each futures market fluctuates dramatically. Therefore, the trend of net total connectedness, the

difference between “to”connectedness and “from”connectedness, is the same as the pattern for connectedness “to”

others. Therefore, the “to”connectedness of each futures market deserves more attention.

Additionally, we use a summation average of the static connectedness of all futures markets, except for their

connectedness to themselves, as the dynamic connectedness of the total network. Total connectedness distills a system into a

single number analogous to total world exports or total world imports. (The two are of course identical.) For example, total

connectedness for the 18 commodity futures equals 69.6% over the full sample; that is, almost two‐thirds of the shocks for the

commodity futures are traceable to the connectedness of shocks across all futures. We find, based on dynamic analysis, that

there is always a high degree of total connectedness in times of turmoil. In 2009 global financial crisis, total connectedness

reaches a peak (93.2%), and after a brief respite, it increases to 90.8% in September 2009. Total connectedness jumps from

79.0% to 94.4% in July 2013, rising by almost 20%, which may be associated with the debt crisis in Europe.

The contribution of our study is empirical rather than methodological. First, using the variance decompositions from

approximating models, we apply a unified framework for the connectedness of global commodities futures at different

levels, from pairwise connectedness to system‐wide connectedness. Second, commodity futures market connectedness

plays an important role in capturing fundamental macroeconomic risks. On the basis of a data set that includes a

2

The International Organization of Securities Commissions (IOSCO) is a global cooperative of securities regulatory agencies that aim to establish and maintain worldwide standards for efficient,

orderly, and fair markets.

XIAO ET AL.

|

599