Erik Adelsohn and Christian Eriksson present their master's thesis

Two approaches to oil price modelling

Abstract

A thesis presented on the dynamics of crude oil prices and the ability
to predict future price and volatility.  We start with a time series
approach using data from 1982 including oil price, gold price,
EUR/USD, JPY/USD, GBP/USD, US commercial oil reserves and VIX
volatility index. We analyze a lot of data using statistical tests
such as regressions, auto-correlation, and cross-auto-correlation. We
discover some interesting connections but find it hard to combine them
into a complete time series model. Furthermore, we take a closer look
at the volatility in oil price and apply models from the GARCH-family.

In the second part we explore the fundamental drives behind the oil
price, namely world production and consumption. We aim at building a
supply-demand model in order to obtain a more stable long term
prediction for the oil price. This approach is easy to understand, but
is in practice hard to calibrate to reality. We research some well
known variables affecting the oil price such as excess capacity,
commercial reserves, substitutes, Peak Oil theories and production
costs. The result is a non-linear supply-demand model. With expert
guidance we create five possible scenarios for oil production
capacity, oil demand and production cost for the year of 2012. These
render five possible future prices. The model is also used in a
powerful Monte Carlo simulation to predict the oil price 2012 based on
normally independently distributed annual growth numbers for capacity,
demand and production cost. In addition to this we discuss other
non-fundamental drives of the oil price such as hedging effects,
foreign exchange rates and price spreads between different types of
oil. Last but not least we give an enlightening chapter about energy
substitutes and compare them to energy from oil.