Multivariate Time Series Modelling of Ex-Pump Prices of Petroleum Products in Ghana

Abstract

The purpose of the study is to obtain a suitable model for the ex-pump prices of petroleum products in Ghana. It examines how changes in the prices of one product cause changes in the price of others in both the short and long terms. Data spanning January 2007 to June 2015 are obtained from the National Petroleum Authority of Ghana, covering four petroleum products; Premium Gasoline, Gas Oil, Kerosene, and Liquefied Petroleum Gas. The analysis is carried out using the technique of Vector Error Corrected (VEC) modelling. This technique is found suitable as the data obtained constitutes a time series that is multivariate in nature, and that the components are found to exhibit long-run relationships. The study reveals that there exists a general upward trend in the ex-pump prices of the products over the period. It also shows that changes in prices of some of the products influence others in the short run, but in the medium to long terms, stability is attained, with Premium Gasoline and Gas Oil prices accounting for most of the variations in changes in prices of the four products. A VEC model of order 1 is found suitable for the data and appears to perform better than the VAR model with the least R-squared of 90.3% for Gasoline prices and the highest of 96% for Gasoil prices. The results reflect the competitive usage of the two products that could serve as substitutes. Thus, to ensure economic prices in petroleum products, in general, it would be expedient to ensure effective management of price changes in these two products.