Commodity momentum and reversal: Do they exist, and if so, why?

• Published date: 01 September 2023
• Author: Meng Han
• Date: 01 September 2023
• DOI: http://doi.org/10.1002/fut.22424

Received: 13 November 2022

|

Accepted: 28 April 2023

DOI: 10.1002/fut.22424

RESEARCH ARTICLE

Commodity momentum and reversal: Do they exist, and if

so, why?

Meng Han

1,2

1

Research Center for Innovative Finance,

Bay Area International Business School,

Beijing Normal University, Zhuhai, China

2

Department of Economics, Econometrics

and Finance, Faculty of Economics and

Business, University of Groningen, the

Netherlands

Correspondence

Meng Han, Research Center for

Innovative Finance, Bay Area

International Business School, Beijing

Normal University, Zhuhai 519087,

China.

Email: m.han0609@hotmail.com

Abstract

Questions as to why differences in momentum and reversal patterns seem to

emerge in commodity futures compared with spot markets, and how these

patterns can be explained, remain unanswered. To investigate these questions,

I examine 23 commodities over a period of 60 years. I first show that including

the net convenience yield in the definition of commodity spot returns

reconciles the differences in the results for commodity spot and futures

markets. Both commodity futures and spot markets exhibit quantitatively

consistent momentum and reversal effects. An initial momentum effect is

followed by a reversal effect and then another momentum effect. These

observed patterns in commodities can be jointly explained by a combination of

traditional asset pricing factors and a basis factor related to the net

convenience yield.

KEYWORDS

asset pricing factors, commodity markets, convenience yield, momentum, reversal

JEL CLASSIFICATION

G10, G11, G13, G14

1|INTRODUCTION

The cross‐sectional momentum effect, known simply as the momentum effect, is one of the most pervasive “anomalies”

in financial literature (see Booth et al., 2016; Conrad & Kaul, 1998; Jegadeesh & Titman, 2001; Koziol & Proelss, 2021).

1

A widely documented momentum and reversal pattern in equity markets is a momentum effect occurring first,

followed by a reversal effect (see Cooper et al., 2004; Jegadeesh & Titman, 2001). However, the momentum and reversal

patterns are not so clear for commodity markets. Nonetheless, understanding these patterns has become increasingly

important due to the financialization of commodity markets (e.g., Cheng & Xiong, 2014).

In this paper, I aim to analyze the momentum and reversal effects in commodity markets and investigate the source

of these effects. Although existing studies have exclusively focused on the momentum and reversal effects in

commodity futures markets, there is still a lack of consensus on these effects among them. Shen et al. (2007) suggest

that in commodity futures markets, an initial momentum effect is followed by a reversal effect when using momentum

strategies with a 2‐month formation period and holding periods of up to 30 months. In contrast, Bianchi et al. (2015)

J Futures Markets. 2023;43:1204–1237.wileyonlinelibrary.com/journal/fut1204

|

© 2023 Wiley Periodicals LLC.

1

Cross‐sectional momentum differs from time‐series momentum. Time‐series momentum focuses on a single asset and decides whether to take a

long or short position in that asset based on its past performance. For instance, a positive past performance indicates a long position in that asset.

This paper reserves the term “momentum”for cross‐sectional momentum.

indicate the presence of a second momentum effect after the reversal effect in commodity futures markets, using

momentum strategies with formation periods of up to 15 months and holding periods of up to 60 months. Chaves and

Viswanathan (2016) represent an important exception to the study of momentum and reversal effects in commodity

spot markets, with their findings suggesting only the presence of the reversal effect.

I argue that the differences in results between commodity futures and spot markets may stem from the fact that

commodity spot returns are often proxied by pure capital gains. However, an appropriate definition of a commodity

spot return should include not only the capital gain but also the net convenience yield, which is a latent payoff of

holding a commodity that is similar to a dividend of a stock (see Pindyck, 1993; Tsvetanov et al., 2016). Ignoring the

role of the net convenience yield in commodity spot returns may influence the observed momentum and reversal

patterns.

Considering the redefined commodity spot return, I comprehensively examine the momentum and reversal effects

in both commodity futures and spot markets. The sample consists of 23 commodities with actively traded futures

contracts, covering the period from September 1960 to September 2020. To explore the momentum and reversal effects

among these commodities, I consider momentum strategies with a broad range of formation and holding periods of up

to 60 months. The results suggest that the inclusion of the net convenience yield in the commodity spot return

definition reconciles the differences in results between commodity spot and futures markets. Specifically, the

momentum and reversal effects coexist and are quantitatively consistent in both commodity futures and spot markets.

The momentum effect emerges first, followed by a reversal effect and then another momentum effect.

The momentum and reversal effects observed in commodity markets differ from those observed in equity markets.

Thus, the behavioral explanation for these effects in equity markets seems incompatible with the effects observed in

commodity markets. Some papers suggest a risk‐return tradeoff to explain the momentum effects in commodity

markets with the basis factor mimicking the risk related to the net convenience yield (see de Groot et al., 2014) and

common asset pricing risk factors, such as those in Fama and French (1993) three‐factor model, Carhart (1997) four‐

factor model and Fama and French (2015) five‐factor model (see Kang & Kwon, 2017).

2

However, these papers

conclude that the risk‐return tradeoff framework seems unable to explain the commodity momentum effects. I argue

that this failure may be because risk premiums are assumed to be constant in the aforementioned studies, although

they vary with time. Therefore, further asset pricing tests are needed to explain the momentum and reversal patterns in

commodity markets.

To jointly explain the momentum and reversal patterns observed in commodity markets, this paper studies the

cross‐sectional returns of different momentum strategies from the perspective of a risk‐return relationship. For

comparisons with existing studies, I consider a one‐factor model with the basis factor, Capital Asset Pricing model

(CAPM), Fama and French (1993) three‐factor model, Carhart (1997) four‐factor model, and Fama and French (2015)

five‐factor model. To study the combined role of the basis factor and the common asset pricing factors, I also consider

augmented models that include the basis factor in the four common asset pricing models. The empirical analyses not

only assume constant risk premiums but also consider time‐varying risk premiums. The results suggest that the model

that combines Carhart (1997) four‐factor model with the basis factor outperforms all other studied models when

accounting for time‐varying risk premiums. Therefore, the momentum and reversal effects observed in commodity

markets can be jointly explained by the risk‐return framework.

To assess the robustness of the findings, several checks are performed in this paper. I construct momentum

strategies in an alternative way, finding that the momentum and reversal patterns in commodity markets and their

risk‐return explanation remain robust. Furthermore, the conclusion that the basis factor in combination with Carhart

(1997) four‐factor model can jointly explain the momentum and reversal patterns in commodity markets is robust to

the particular selection of momentum strategies. Additionally, I repeat the analysis with different estimation methods,

including rolling window betas, and find that the results remain robust. Moreover, I repeat the analysis with relative

basis as an alternative factor that mimics the risk related to the net convenience yield, and obtain similar results.

Finally, I consider other commodity‐specific factors that have been shown to successfully explain commodity returns,

such as the basis‐momentum, skewness, convenience yield risk, hedging pressure, and speculative pressure factors.

The common asset pricing factors, such as the equity momentum and profitability factors, remain significantly priced

even after controlling for these commodity‐specific factors.

2

The basis is commonly defined as the time‐scaled difference between the logarithms of the spot price and the futures price.

HAN

|

1205

This paper contributes to the existing literature on commodity momentum and reversal effects by arriving at a

comprehensive understanding of the effects in commodity markets. In addition to commodity futures markets, I extend

the discussion to commodity spot markets for three reasons. First, physical commodities underlie futures contracts,

meaning that the commodity spot and futures prices are closely linked. Thus, the momentum and reversal effects in

spot markets, reflecting the relative spot price behavior among commodities, are also essential for investors in futures

markets. Second, investors may engage in spot markets to make profits if the difference between the futures and spot

prices of a commodity is sufficiently large to cover relevant storage costs and interest (see Tilton et al., 2011). Third,

producers and consumers of commodities are subject to fluctuations in production and/or consumption, and therefore,

hold physical commodities (see Pindyck, 2001). This highlights the importance of commodity spot markets. However,

the momentum and reversal effects in these markets remain relatively unexplored in the literature. I add to this

literature by providing novel evidence for spot markets in showing that once the definition of commodity spot return

includes the net convenience yield, the momentum and reversal effects in spot markets are quantitatively similar to

those in futures markets.

This paper also contributes to the existing literature on momentum and reversal effects by providing a novel

perspective to explore their sources using a risk‐return framework. I add to this literature by providing evidence that

the returns from commodity momentum strategies compensate for risks, such as the size and equity momentum

factors. This finding provides new insights into the understanding of momentum and/or reversal effects in other assets,

such as stocks and bonds. Specifically, the momentum “anomaly”observed in other assets could potentially be

explained through a risk‐return framework.

The rest of this paper is organized as follows. Section 2explains the adopted commodity return definitions,

commodity momentum strategies, and asset pricing tests. Section 3describes the commodity data and risk factors.

Section 4presents and discusses the results regarding the performance of commodity momentum strategies, risk‐return

explanation for momentum and reversal patterns with both constant and time‐varying risk premiums, and robustness

and sensitivity checks. Section 5concludes the paper.

2|METHODOLOGY

2.1 |Defining commodity returns

This paper assumes that investors in futures markets earn fully collateralized futures returns, consistent with findings

from previous studies, such as Bianchi et al. (2016), de Groot et al. (2014), Fuertes et al. (2010), Miffre and Rallis (2007),

and Paschke et al. (2020). When investors open a long position in a futures contract, they do not make any payments;

therefore the monetary amount equivalent to the value of the traded futures contract is assumed to be invested in a

“safe”asset, such as a treasury bill or a margin account. This means that investors are assumed to engage in an

unleveraged position in futures.

3

As such, the fully collateralized futures return is comparable to a stock or commodity

spot return. Consider a futures strategy in which at time

t

, an investor opens a long position in a futures contract

maturing two periods later, specifically at time

t

+

2

. To clarify, in the following empirical analysis, each period refers

to 2 months (see details in Section 3). This investor ends the position at time

t

+

1

(2 months later) and subsequently

rolls over to the next futures contract maturing at

t

+3

. The fully collateralized futures return for this strategy from

time

t

to

t

+

1

(denoted as

r

fu tt,+

1

→

or

r

fu t,+

1

) is defined as

rF

Frf=+−1

,

fu tt

tt

tt tt

,+1

+1, +2

,+2 +1

→→(1)

where

Ftt,+

2

and Ftt+1, +

2

are the futures prices at time

t

and

t

+

1

of a futures contract maturing at time

t

+

2

.

rf

tt+

1

→

(

rf

t

) refers to the net collateral return on the risk‐free asset. This paper calculates futures return with prices of the same

futures contract, which makes the return replicable for investors. This approach is a common practice in commodity

literature (see Chaves & Viswanathan, 2016; Kang & Kwon, 2017; Paschke et al., 2020; Shen et al., 2007).

3

It is noteworthy that in practice, investors often engage in leveraged position in futures to achieve higher returns. Therefore, the futures return on

fully collateral basis used in this paper is more conservative.

1206

|

HAN