A multi-resolution and multivariate analysis of the dynamic relationships between crude oil and petroleum-product prices

Abstract

This paper proposes the use of a novel multivariate, dynamic approach wavelet local multiple correlation (WLMC) (Fernández-Macho, 2018) to analyse the relationship between oil time series in the time-scale domain. This approach is suitable for use with energy data of any kind that change over time and involve heterogeneous agents who make decisions across different time horizons and operate on different time scales. The study of the links between multivariate oil time series is of great importance in energy research, e.g., it is extremely important for petroleum industry refiners and investors to know the relationships and margins between output prices and crude oil costs. The estimation of wavelet correlations in a multivariate framework between such prices is a suitable way to analyse crude oil and petroleum products as a system. To exemplify the use of WLMC, we analyse the relationships between the prices of seven commodities: West Texas Intermediate crude oil and six distilled products (conventional gasoline, regular gasoline, heating oil, diesel fuel, kerosene and propane) from 10/06/2006 to 17/01/2017. The results reveal that the wavelet correlations are strong throughout the period studied and there is a strong decay in correlation values from 2013 to 2015. The most plausible explanation for this decay is overproduction of tight oil in the U.S. and a slowdown in global demand for oil. Furthermore, our results also reveal that heating oil, diesel and kerosene maximise the multiple correlation with respect to the other oil variables on different scales, indicating that these products are the most dependent variables in the crude-product/price system. WLMC offers new opportunities for applications in energy research and other fields. © 2018 Elsevier Ltd