The time-varying causality between spot and futures crude oil prices: A regime switching approach

Abstract

One puzzling result in the literature on the linkages between the spot and futures crude oil prices is the sensitivity of causality tests to the sample period that one considers, which is usually handled by sample splitting. In order to overcome this difficulty, we propose a model that allows time-varying Granger causality between these prices. The model is used to investigate the time-varying causal linkages between the daily spot and futures prices for maturities of one, two, three and four months of the West Texas Intermediate (WTI) crude oil benchmark over the period January 2, 1986-July 31, 2013. The results indicate that the causal links between these oil prices are strongly time-varying. Both variables have a predictive power for each other during various subperiods, but not in all periods. Further, these periods coincide with major influential changes in the oil and stock markets and the geopolitics, implying that the findings are not statistical artifacts, but reflect real economic, financial and geopolitical regime changes. The full sample's conditional Granger causality tests based on the Markov-switching vector-error correction (MS-VEC) model reject both the causal impact of the lagged futures prices on the spot prices and the causal impact from the lagged spot prices on the futures prices. Therefore, the results show that the lead-lag relationship between the spot and futures oil markets exists only temporarily. While encompassing the previous findings, they also offer new insights into the nature of the lead-lag relationships between the spot and futures oil markets by taking non-linearity and time-variation into account. (C) 2015 Elsevier Inc. All rights reserved.