Jump Processes in the Market for Crude Oil

In many commodity markets, the arrival of new information leads to unexpectedly rapid changes – or jumps – in commodity prices. Such arrivals suggest the assumption that log-return relatives are normally distributed may not hold. This article investigates the potential presence of such jumps in the price of crude oil, both in terms of spot prices and the futures prices. The investigation is carried out over three data frequencies (Monthly, Weekly, Daily), which allows for an investigation of temporal properties. Based on the methodology employed, likelihood ratio tests are used to compare among four stochastic data-generating processes. Maximum-likelihood estimation results suggest that jumps are important when examining high frequency (Daily) data, but the significance of jumps is ‘washed out’ at lower frequencies (Monthly). However, allowing for time-varying volatility calls into question the empirical relevance of jumps, even for higher frequency data.