Predicting Commodity‐futures Basis Factor Return by Basis Spread

• Author: Daehwan Kim
• DOI: http://doi.org/10.1111/ajfs.12100
• Date: 01 August 2015
• Published date: 01 August 2015

Predicting Commodity-futures Basis Factor

Return by Basis Spread*

Daehwan Kim**

Department of Economics, Konkuk University

Received 20 December 2014; Accepted 25 May 2015

Abstract

A growing body of literature confirms the significance of the commodity futures basis factor.

It has a significantly positive premium and it explains the cross-section of commodity-futures

excess returns. We extend the literature by documenting the predictive relation between this

factor and the inter-quartile spread in the basis; the predictability of the basis factor return

has not been previously reported. From the simple regression analysis of the historical com-

modity futures data we show that the basis spread is a strong predictor of the basis factor

return. We discuss the implication of this finding on the nature of the basis factor; we also

discuss the market timing strategies based on the basis spread.

Keywords Commodity futures; Basis factor; Basis spread; Predictability; Market timing strate-

gies

JEL Classification: G10, G12, G13

1. Introduction

An increase in futures price is an excess return to the holder of the futures contract

since holding a futures contract does not require any equity capital commitment. In

the commodity futures market, the excess return varies across commodities, and

explaining the cross-section of the excess return has been an active research agenda

in recent years. A growing body of literature has shown that the basis factor has an

important role in this regard. The basis is the ratio of the spot price to the futures

*

The paper has benefited from useful comments by Sangwon Suh, Kyuhyong Kim, Wenjie

Ma, Ke Tang, Jinho Bae, Jeffrey Nilsen, and participants of the 2012 Korea International Eco-

nomic Association Conference, 2013 Joint Conference of Finance Associations of Korea, 2013

Asia Pacific Association of Derivatives Conference in Busan, and the 2nd International Con-

ference on Futures and Derivative Market in Beijing. This work was supported by the

National Research Foundation of Korea Grant funded by the Korean Government (NRF-

2012S1A5A8024387).

**Corresponding author: Daehwan Kim, Department of Economics, Konkuk University,

Gwangjingu, Seoul, 143-701, South Korea. Tel: +82-2-450-3937, Fax: +82-2-450-3615,

email: dkim@konkuk.ac.kr.

Asia-Pacific Journal of Financial Studies (2015) 44, 587–615 doi:10.1111/ajfs.12100

©2015 Korean Securities Association 587

price, and the basis factor return is calculated as the return difference between two

portfolios of high- and low-basis commodity futures. Gorton and Rouwenhorst

(2006), Fuertes et al. (2010), and Gorton et al. (2012) report a significantly positive

premium accruing to the basis factor. De Roon and Szymanowska (2010), Yang

(2013), Szymanowska et al. (2013), and Bakshi et al. (2014) report that the beta of

the basis factor explains a significant part of the cross-section of commodity-futures

excess returns.

1

We extend the literature by documenting the predictability of the basis factor.

As far as we know, the predictability of the basis factor return has not been previ-

ously reported and our paper is the first in this regard. We calculate the inter-quar-

tile spread in the basis from the cross-section of commodities, and show that this

spread predicts the basis factor return.

2

When the next-period basis factor return

(y) is regressed on the current-period basis spread (x), the coefficient estimate is

statistically significant.

Our predictive regression equation can be motivated from positive cross-sec-

tional correlation between the basis and the next-period excess return. We may

apply the “inter-quartile-spread operator” to each variable and obtain positive rela-

tionship between the basis spread and the spread in the next-period excess return.

The excess return spread is essentially identical to the basis factor return. Thus, a

positive relationship will also exist between the basis spread and the basis factor

return. That the return to a characteristics-sorted long-short portfolio (such as the

basis factor) can be predicted by the spread in the characteristics (such as basis

spread) is a well-known idea in finance. This idea has been successfully applied to

the prediction of stock-market factors, such as the value factor and the momentum

factor. See, for example, Asness et al. (2000), Cohen et al. (2003), and Stivers and

Sun (2010).

While the spread-predict-the-factor-return idea is straightforward, confirming

this idea in the commodity futures data is not a trivial exercise. Three important

implications can be drawn from our predictability results. First, our predictability

results support the view of Daniel and Titman (1997) that it is not just beta,

but also characteristics (such as basis spread) that matter. If we take the ‘only-beta-

matters’ perspective of Davis et al. (2000), the time variation in the basis spread

1

Szymanowska et al. (2013) use the term “the basis factor.” Yang (2013) calls it “the slope

factor,” and Bakshi et al. (2014) refer to it as “the carry factor.” As noted by Szymanowska

et al. (2013), the commodity-futures basis factor is comparable to the foreign-exchange-mar-

ket carry factor; thus, the carry factor is certainly another suitable name for the basis factor.

2

We calculate the inter-quartile spread in the inverse basis rather than in the basis, as the for-

mer is a more accurate representation of the theoretical relationship that we discuss later.

Thus, the proper name for our spread variable should be the inverse basis spread. As it

sounds rather clumsy, however, we call our spread variable simply the basis spread. Note also

that our usage of this term is different from Yang’s (2013). In Yang’s paper, the basis spread

refers to the premium accruing to the basis factor.

D. Kim

588 ©2015 Korean Securities Association