Volatility term structures in commodity markets

• Author: Marcel Prokopczuk,Christoph Würsig,Fabian Hollstein
• Published date: 01 April 2020
• DOI: http://doi.org/10.1002/fut.22083
• Date: 01 April 2020

© 2019 The Authors. The Journal of Futures Markets published by Wiley Periodicals, Inc.

J Futures Markets. 2020;40:527–555. wileyonlinelibrary.com/journal/fut

|

527

Received: 13 September 2019

|

Accepted: 22 November 2019

DOI: 10.1002/fut.22083

RESEARCH ARTICLE

Volatility term structures in commodity markets

Fabian Hollstein

1

|

Marcel Prokopczuk

1,2

|

Christoph Würsig

1

1

School of Economics and Management,

Institute for Financial Markets, Gottfried

Wilhelm Leibniz University of Hanover,

Hannover, Germany

2

ICMA Centre, Henley Business School,

University of Reading, Reading, UK

Correspondence

Christoph Würsig, School of Economics

and Management, Gottfried Wilhelm

Leibniz University of Hanover,

Koenigsworther Platz 1, 30167 Hannover,

Germany.

Email: wuersig@fmt.uni-hannover.de

Abstract

In this study, we comprehensively examine the volatility term structures in

commodity markets. We model state‐dependent spillovers in principal

components (PCs) of the volatility term structures of different commodities,

as well as that of the equity market. We detect strong economic links and a

substantial interconnectedness of the volatility term structures of commodities.

Accounting for intra‐commodity‐market spillovers significantly improves out‐

of‐sample forecasts of the components of the volatility term structure. Spillovers

following macroeconomic news announcements account for a large proportion

of this forecast power. There thus seems to be substantial information

transmission between different commodity markets.

KEYWORDS

commodities, information transmission, spillovers, volatility term structure

JEL CLASSIFICATION

G10; G14; G17

1

|

INTRODUCTION

A large set of external events and conditions has the potential to affect commodity markets. Important drivers of

commodity prices are, inter alia, weather, investor flows, and macroeconomic conditions. While the level of commodity

prices is certainly important, understanding the volatility of commodity prices is at least as crucial. For example,

Pindyck (2004) shows that, because storage helps to smooth production and deliveries, the marginal value of storage

increases with volatility. Further applications where volatility is of special concern include risk management decisions,

margin calculations, or the valuation of options contracts. While previous studies have examined the impact of

commodity spot volatility, the entire volatility term structure provides additional important information for the above

mentioned issues, since short‐term and long‐term options embed partly differential information and provide market

expectations of future volatility over various horizons.

Theimportanceofconsideringtheentiretermstructurehas been widely documented for equity markets (e.g., Adrian &

Rosenberg, 2008; Bakshi, Panayotov, & Skoulakis, 2011; Feunou, Fontaine, Taamouti, & Tédongap, 2013). In particular,

these studies show that the volatility term structure is informative about, inter alia, risk premia, measures of real economic

activity, business cycle risk, and the tightness of financial constraints. Investigating the interconnectedness of the term

structure and its relation with macroeconomic variables and announcements can be crucial to help understand the

interdependencies and macroeconomic links of the commodity markets. This can be particularly helpful for practitioners

that can use predictability of the entire volatility term structure for more accurate risk evaluations of their portfolios.

-----------------------------------------------------------------------------------------------

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided

the original work is properly cited.

Our main contribution is to provide a comprehensive study of the volatility term structure of different commodity

markets. The volatility term structure is of special interest for commodity markets because of its relation with the so

called Samuelson (1965) effect. This effect states that volatility generally decreases with increasing time to maturity. In

appreciating this, we can enhance our understanding of the determinants and dynamics of the volatility term structure.

First, we decompose the volatility term structure into its principal components (PCs) and study their economic

drivers. We focus on the first three PCs: the level, the slope and the curvature of the term structure. This analysis allows

us to understand how volatility dynamics change for contracts with different expiry dates.

When we investigate the macroeconomic determinants of the commodity volatility term structure, we uncover two main

results. (a) Macroeconomic variables can explain a large proportion of the variation in the level factor, and typically a

somewhat smaller share for the slope and curvature factors. (b) An increase in the proportion of speculative open interest

reduces the volatility level for various markets, while employment is positively related to the volatility level.

Second, we use a state‐dependent autoregressive (AR) model to examine volatility spillovers between commodity

markets. We compare a model using only the past lags of one commodity volatility term structure to a state‐dependent

unrestricted AR‐model which also includes the lagged volatility PCs of another commodity, following the causality

model by Granger (1969, 1988). We define economic states based on the forecast of the Engle and Manganelli (2004)

conditional autoregressive Value at Risk (CaViaR). Using the Granger (1969, 1988) causality model to make out‐of‐

sample predictions of the implied volatility term structure generally yields sizable forecast improvements over the

predictions of the simple state‐dependent AR‐model. Accounting for spillover effects for the level and the slope yields

out‐of‐sample R

2

s of up to 5%. Intra‐commodity effects are more important for the commodity market than spillover

effects originating from the equity market. Finally, spillovers are state‐dependent: they are strongest during market

distress and smallest during normal periods.

One possible explanation for these findings is information transmission. To isolate the effects originating from this

channel, we investigate the impact of scheduled macroeconomic news announcements on spillovers. If spillovers are

larger after macroeconomic news announcements, this would indicate that some commodity markets capture

information on macroeconomic news earlier than others. This could lead to subsequent changes in the volatility term

structure of cross‐section of the commodity markets. We find that macroeconomic news announcements models do

indeed explain up to 70% of the spillovers for the level. News announcements associated with consumer income or

consumer sentiment have a particularly large influence on spillovers for all components of the term structure.

We also investigate the impact of the financialization of commodity markets, which leads to a stronger comovement

across commodities in recent years due to the increased use of commodities as an investment (Christoffersen, Lunde, &

Olesen, 2019; Tang & Xiong, 2012). We conduct a subsample analysis by studying changes in the lead/lag relationship

between commodity markets pre‐and post‐financialization, which reveals two main findings: First, the volatility term

structure for commodity and equity markets is strongly integrated for the post‐financialization period. Second, there are

two effects that affect spillovers postfinancialization: (a) the increase in contemporaneous movements lowers spillovers

for the level and (b) more common factors for the slope and the curvature lead to overall higher spillovers.

Our study is related to several strands of the literature. For equity and bond markets a variety of articles show that the

variance term structure is important and can capture unobserved risk factors. Adrian and Rosenberg (2008) and Bakshi et al.

(2011) show that factors that describe the volatility term structure can predict various economic and financial measures.

Bakshi et al. (2011) draw on an analogy with the term structure of interest rates and argue that the variance term structure

embodies expected variances by both the financial and the real sector, as perceived by the index option market.

1

For commodity markets, there is a vast literature that finds a factor structure in returns. Rotemberg and Pindyck

(1990), Yang (2013), Szymanowska, De Roon, Nijman, and Van Den Goorbergh (2014), and Bakshi, Gao, and Rossi

(2017) argue that common factors in commodity markets can explain a large proportion of cross‐sectional return

variation. For their analyses, these studies use the cross‐section of commodity returns. Brunetti, Büyükşahin, and

Harris (2016) show that hedge funds positions are negatively related to the volatility in corn, crude oil, and natural gas

futures markets. Hammoudeh and Yuan (2008) investigate the effects of oil and interest rate shocks on the volatility of

metals markets, using various GARCH model specifications.

Our study extends this literature by investigating the entire volatility term structure for a large cross‐section of

commodity markets. Leveraging the various expiration dates of commodity futures and options enables us to study the

term structure and analyze whether there is a common factor structure in the volatility term structure.

1

Further studies on the volatility term structure in equity markets include: Campa and Chang (1995), Mixon (2007), Johnson (2017), and Hollstein, Prokopczuk, and WeseSimen (2019).

528

|

HOLLSTEIN ET AL.