Econometric Modelling of Oil Prices

Chapter 1
Introduction
The importance of oil is well known to everyone. It is one of the most
important source of energy and it is of critical importance to the maintenance
of the industrialised civilisation. Economics, politics and technology are all
deeply aected even by slight variations of the price of the crude. According
to the EIA
1
its importance has been decreasing over time especially in the
last thirdy years, but it still represents the largest energy source with a 35%
share over the world total energy demand.
Oil market dynamics has therefore a huge impact on all economic activi-
ties. Various transmission channels exist through which oil prices may have
an impact on economic activity. A higher crude oil price means an higher
energy bill for the private consumer and an higher production cost for the
industrial producers. Therefore productivity may be aected, causing conse-
quences for wages, employment, ination, prots and investments as well as
stock market capitalization
2
.
From the supply side, higher prices mean reduced availability af an input
of production. The consequences are reduced output growth and increasing
costs. In this way productivity is slowed.
Inationary shocks which are likely to be accompanied by second round
eects are often the result of an abnormal increase in oil price. The eects to
1
The Energy Information Agency provides the ocial statistics for the U.S. government
2
Market capitalization: is a measurement of size of a business enterprise equal to the
share price times the number of shares outstanding of a publicly traded company. Capi-
talization could represent the public opinion of a company net worth and is a determining
factor in stock valuation. Likewise, the capitalization of stock markets may be compared
to other economic indicators (Wikipedia, Market Capitalization)
9
the economy can then be long lasting causing vicious circles
3
and inationary
expectations which may aect consumption, investments and employment,
altering the structure and the allocation of factors of production inside the
economies.
For this reasons, energy producers and consumers as well as central gov-
ernments try to forecast the trend the price will follow over long time hori-
zons in order to plan strategies, evaluate investment opportunities and select
the timing for policy implementation. Considering, for example, that an oil
drilling project to be carried out requires 20 to 30 years, it is easy to under-
stand why it would be of critical importance to forecast trends in order to
evaluate the protability of the project itself.
In order to make forecast it would be intuitive to design a structural
framework in which the fundamentals of supply and demand are analyzed.
However, this can represent a dicult challenge since not all the explanatory
variables, such as investments and production capacity, inventories and the
key drivers of demand, are easy to anticipate. In addition, politics plays an
important role in the market. Therefore, modelling a structural framework
could turn out to be too challenging to perform. An alternative is to refer to
certain stochastic properties of known processes to try to forecast a move-
ment that is consistent with the oil price behavior. The stochastic properties
of crude oil prices have, in fact, important implication for forecasting. The
choice of a stochastic process to represent the evolution of the series is crucial
to measure the impact of uncertainties over the decision to invest. It is nec-
essary to choose a movement that reects as likely as possible the dynamics
of the world oil market.
The issue of whether it is preferable to choose a certain stochastic process
is important because investments are irreversible (at least in part) and, as
such have option like characteristics
4
. An investment decision based on a
mean reverting process could turn out to be quite dierent from one based
on a random walk. As Pindyck (1999, The Long{Run Evolution of Energy
Prices) pointed out, the former choice would be consistent with a model
3
Such as price{wage loops: higher wages are demand by workers to keep up with
ination and higher prices are set by rms in order to get a certain amount of prots. In
turn, higher wages are demanded and this creates the vicious circle.
4
When a rm makes an irreversible investment expenditure, it exercises its option to
invest. It gives up the possibility of waiting for new information to arrive that might aect
the desirability or timing of the expenditure (Dixit & Pindyck(1994), Investment Under
Uncertainty)
10
of a competitive market in which the price should revert towards long run
marginal costs, which is likely to change only slowly. This would imply that
in the long run the process will be dominated by a particular trend rather
than following a random walk, and that price shocks eects will be only
temporary.
The stochastic properties of oil prices are crucial for policy makers too.
It has been argued in fact, to what extent monetary policy, as opposed to
oil price shocks, contributed to the stagation following the crisis of the
1970s. The argument is that policy makers responded to the inationary
pressures caused by price shocks by raising interest rates, thereby causing
a deep recession that would not have occurred without the central bank’s
interventions. Now, given the policy targets set by actual central banks, it
would be interesting to analyze how those pressures could be dealt with in
order to avoid similar recessionary eects in the future.
This work, based on the paper "The Price of Oil over the Very Long
Term" by Sophie Cardon included in The Econometrics of Energy Systems
(2007, Palgrave Macmillan), is a univariate analysis in which the long run
behavior of crude oil prices is under examination. Two methods of dealing
with the data have been used here. The rst is based on unit root proce-
dures and OLS estimation while the second is based on state space modelling
and Kalman ltering. The analysis is structured in 5 Chapters. After the
introduction, an overview of the oil industry and its evolution over time as
well as the descriptive statistics of the WTI series is presented in Chapter
2. Chapter 3 is devoted to unit root testing where the classical approach in
testing for stationarity is compared with the results of tests allowing for a
structural break in the series, rst developed by Perron. One of these lat-
ter specications allows the rejection of the unit root hypothesis, indicating
mean reverting features in a time-varying trend model. Chapter 4 is de-
voted to the description of the state space approach that enables the analyst
to design a trend reverting model in which the trend may vary over time.
An Ornstein{Uhlenbeck process has been considered and approximated in
discrete time in order to be estimated. The Kalman lter is being used to
estimate the market dynamics and to produce forecasts; estimation results
as well as model checking will precede the forecasting procedure in which the
data are cast in two dierent specications of the state space system to be
compared. It will be seen how the OLS, from the unit root tests, performs
dierently from the Kalman lter estimation; this latter one still enables the
analyst to consider the series as a mean reverting process to a time{varying
11
trend but with a higher speed of reversion. Forecasting procedures extends
the vector of observations until 2020. Such operation turns useful in making
future investment decisions and in designing strategic policies. Concluding
remarks and a summary of results are provided in Chapter 5.
12
Chapter 2
Overview of the Oil Industry
The petroleum industry includes the global process of exploration, extraction,
rening, transporting and marketing petroleum products, mainly fuel oil and
gasoline(petrol)
1
. Taken as a whole, the industry is the largest in terms of
dollars value.
It can be divided into two main components: upstream and downstream.
The upstream sector includes the exploration and production sector which
refers to all the activities required to search and recover crude oil. The
downstream sector, instead, includes all the rening activities as well as the
selling and the distribution of the oil products.
It can be dicult to describe the petroleum industry solely in economic
terms. In addition to the terms imperfect competition and economic rent,
it is necessary to consider the political sphere that inuenced the market’s
framework. The petroleum industry has changed over time according to dif-
ferent regulations and dierent regulation authorities. The oil issue has been
an important matter for foreign policy actions that many times concerned
wars over the possession of reserves. Therefore, to get a clear understand-
ing of the oil price dynamics it is important to t the discussion inside a
framework in which the international oil market is examined. In fact, oil
history starts more than 150 years ago with the discovery of the resource by
Colonel Drake and William A. Smith in 1859 and, since then, the industry
has undergone dramatical developments. From the oil rush and the intense
competition in America by fortune seekers in the second half of the 19th
century to the OPEC era, oil prices have evolved according to dierent fun-
1
Wikipedia, Petroleum Industry
13
damentals which have led the commodity to increase dramatically in the last
fourty years, while they have been considerably low for most of the previous
oil history.
2.1 The Early Stages (1859 - 1960)
After the early discovery, oil was basically used for limited lighting purposes
and the industry was mainly an American industry where the law allowed
landowners to have rights on the underground resources. There was a frenzy
to buy land and pump out oil from the earth and every time a new eld
was discovered prices felt and output soared leading to excessive drilling
and wastage. Prices varied a lot during this period. Figure 2.1 depicts
the evolution of oil prices since 1860 and the behavior of the series clearly
represents any oil market development.
In 1863 then, a new subject entered the scene aimed to control the entire
industry. This was John D. Rockefeller whose strategy was to control the
bottlenecks facilities in order to dominate the industry. Controlling rening,
transportation and the distribution segments meant the control of the whole
oil activity in America. He and his assistant H. Flagler gave birth to the
Standard Oil Company in 1870 which would have controlled the market for
several years through the exploitation of economies of scale, being able to
inuence prices using its market power. However a series of events would
have threatened Standard Oil dominance in later years.
Oil was found in other countries, especially in Russia where an oil pro-
duction increased rapidly during the 1870s. Operations in Indonesia was
then carried out by the Royal Dutch Company and Shell which later on de-
cided to merge together in order to pursue their global ambitions; in the rst
decade of the 20th century, the Royal Dutch/Shell was founded and started
operations in east{Europe, south America, Russia, US and North Africa. In
addition there were large discoveries in Texas in 1901 and the emergence of
two other important companies such as the Texas Oil Company and the Gulf
Oil Company. However, the most important fact the threatened Standard
Oil was the passage of the Sherman Antitrust Act in 1890
2
which would have
later led to the division of the Rockefeller’s company into separate entities.
2
The Sherman Antitrust Act was the rst regulation in America which limited the
power of cartels and monopolies, and it still is the basis for the today’s American antitrust
law.
14