Oil Price Modelling

In 2008 Incremental worked with McCormick Rankin Cagney on a project for the Auckland Regional Council modelling oil prices out to 2060.

The model starts with an estimate of total available oil, and the price bands in which it is currently available, and an estimate of oil demand in each price band. Each simulated year, the market is cleared by matching supply and demand stacks, and an oil price and volume is determined. At the end of each year, the availability and demand for oil are modified as a result of the price - a higher price means less demand in the future, but greater production effort into the shrinking available oil. Lower prices mean higher demand, but less production.

The relationships between price and supply and demand are derived from historical data, and are not linear. As a result, the model displays second-order behaviour, with price trajectories exhibiting peaks and troughs.

Any single forecast over such a long period will certainly be wrong, so we use Monte-Carlo simulation to generate thousands of oil price and usage trajectories and generate price and usage bounds (and a mean price) for each year.

A sample result for oil price from the model is shown below, indicating 95% confidence bounds (bars), a mean trajectory (blue), a median trajectory (magenta) and 10 sample individual trajectories (light grey). Results from the model are surprisingly robust to one's beliefs about available reserves and the shape of the demand stack.

Following the client's preferences, the model is implemented as an Excel spreadsheet. The Monte-Carlo simulation is implemented in Visual Basic.

The full report is available from the ARC here.