Forecasting OPEC Oil Price: A Comparison of Parametric Stochastic Models

Abstract

Most academic papers on oil price forecasting have frequently focused on the use of WTI and European Brent oil price series with little focus on other equally important international oil price benchmarks such as the OPEC Reference Basket (ORB). The ORB is a weighted average of 11-member countries crude streams weighted according to production and exports to the main markets. This paper compares the forecasting accuracy of four stochastic processes and four univariate random walk models using daily data of OPEC Reference Basket series. The study finds that the random walk univariate model outperforms the other stochastic processes. An element of uncertainty was introduced into the point estimates by deriving probability distribution that describes the possible price paths on a given day and their likelihood of occurrence. This will help decision makers, traders and analysts to have a better understanding of the possible daily prices that could occur.

JEL Classification Numbers: E64; C22; Q30

Keywords: Oil Price Forecasting, Probability Distributions, and Forecast Evaluation Statistics, Brownian Motion with Mean Reversion process, GARCH Models