The role of financial speculation in the world crude oil market: TVP-VAR and BVAR-SV approaches
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
When the crude oil price rocketed to $147 per barrel in July 2008 and then
dropped to as low as $30 per barrel in December 2008, it catalyzed a hot debate about
the factors of oil price fluctuations. A large number of papers argue that the main
driver of the oil price fluctuations from 2003 to 2008 was due to economic
fundamentals in the form of rapidly growing oil demand with stagnant oil supply.
However, a different view is that speculation in the oil futures market caused the oil
price to drift away from the level justified by the fundamental market forces of
demand and supply because a large amount of investment flowed to the oil futures
market during this period. This dissertation links the oil financial and spot markets
through the oil futures-spot price spread and investigates if the financial activity in the
oil futures market plays a critical role in oil spot price fluctuations between 2003 and
2008. In addition, this dissertation also discusses the recent oil price drop since July
2014 and studies whether the main driver of this recent oil price change is similar to
that of the oil price change in 2008. ☐ Unlike other related literature that uses standard structural VAR, this
dissertation applies a Time Varying Parameter Vector Autoregression (TVP-VAR)
model with stochastic volatilities that can capture both time-varying relationships
between economic aggregates and time-varying impacts of different oil shocks. This
approach disentangles the oil financial speculation shock from economic fundamental
shocks. In the meantime, the findings of the TVP-VAR model are compared with
those of the Bayesian VAR with stochastic volatilities (BVAR-SV) model, a
benchmark model in this dissertation, to see if incorporating time-varying coefficients
in the model can give better results. The results of the comparison show that the time
variations in coefficients are insignificant and imposing time varying coefficients in
the model not only increases the estimation computation work load but also affects the
model’s estimation accuracy. Therefore, the conclusion in this dissertation comes from
the results of the BVAR-SV model. The results imply that the large proportion of the
oil price changes from 2003 to 2008 can be explained by the oil demand shock but this
proportion has been decreasing since 2005. In addition, the contribution of the oil
financial speculation shock has increased substantially in recent years. In sum, the
main driver of oil price change is oil demand from 2003 to 2008, whereas the main
driver from 2014 to 2015 is oil financial speculation in the oil futures market.