Explaining the oil price spread between WTI and Brent during U.S.

shale oil revolution

Dimitrios Kontaxis Master ThesisMEL Class of 2015-2016

September 12, 2016

OutlineI. Introduction

II. Literature Review

III.Methodology

IV. Main Results

V. Conclusions

Introduction

Figure: Crude oil spot prices Source: Thomson Reuters

• WTI and Brent oil prices were moving in tandem until late 2010 with WTI trading at a slight premium.

• WTI started decreasing in level with Brent not following the same downward movement.

• From 2014 and onwards the two crude oil prices seem to move together again with WTI trading at a slight discount compared to Brent.

Research Objective: Explore the properties of the relationship between the two oil benchmarks by analyzing the structure and the underlying drivers of WTI-Brent spread movement.

Research Question: What is the relationship of the oil price spread from 2010 until 2016 and which factors can explain this relationship?

OutlineI. Introduction

II. Literature Review

III.Methodology

IV. Main Results

V. Conclusions

Literature Review• Until 2010, there was long-term link between the two crude oil prices given by the following

equation:

• Price spread WTI-Brent experienced a structural change on 15 December 2010. The oil price spread moved from a stationary to a non stationary process.

• Events that affected the prices independently:

WTI BRENT

• U.S. shale oil revolution• Increasing Canadian oil imports to

PADD2• Capacity constraints in the

transportation infrastructure• U.S. Refinery Configuration• Inability to export crude oil

• Arab Oil Spring• Fukushima accident

Outline

I. Introduction

II. Literature Review

III.Methodology

IV. Main Results

V. Conclusions

Methodology• Chow test for structural break: The Chow test is an F test which shows whether time series

data from two different time periods could show the same relationship or not. The Chow test can only be used when someone has a priori idea of the date that the coefficients of a time series have changed between the two sub-samples defined by the break date. This test can be applied to test whether a major shock in the oil market has changed the spread relationship.

• Unit Root tests for (non)stationarity: The main difference between stationary and non stationary time series lies to the fact that when random shocks happen, the first one is mean reverting, meaning that there is a tendency to return to its long-term mean, while the shocks in the second one have a persistent effect causing the time series to be in the so called random walk.

• Auto Regressive Distributed Lag model: The use of the ARDL model helps to test for long-term relationships between core variables without ending to spurious regressions since the long run coefficients can be consistent when you associate stationary with non stationary time series data or even when you regress non stationary variables. As such the form of our general model will be:

Outline

I. Introduction

II. Literature Review

III.Methodology

IV. Main Results

V. Conclusions

Main Results

Hypothesis: The WTI-Brent spread experienced a structural break in the first quarter of 2014.

Figure: Monthly oil movements from PADD2 to PADD1 and PADD3 Source: U.S. EIA

The oil price spread of WTI-Brent experienced a structural break on 13 March 2014.

Chow test for structural break

Keystone XL pipeline expansion

Main ResultsBased on the break date on 13 March 2014, we examine the process of the price spread with unit root tests:

• From 15 December 2010 until 13 March 2014, the oil price spread follows a non stationary process.

• From 14 March 2014 until 31 May 2016, the oil price spread has formed a new stationary relationship with WTI trading at a slight discount compared to Brent.

Main ResultsBased on the ARDL model, we came up with the following results:

1. The oil price spread WTI-Brent has a positive relationship with the oil pipeline flows from PADD2 to PADD3.

2. The oil price spread WTI-Brent has a negative relationship with the number of U.S. operational oil rigs.

3. The oil price spread WTI-Brent has a negative relationship with the Canadian oil imports to PADD2.

Outline

I. Introduction

II. Literature Review

III.Methodology

IV. Main Results

V. Conclusions

Conclusions• Large discounts in the WTI price caused the spread to significantly decrease in 2011 with

researchers finding a structural change in the relationship between WTI and Brent on 15 December 2010.

• The oil price spread formed a non stationary relationship.

• U.S. shale oil revolution along with increasing Canadian oil imports to PADD2 substantially increased the storage levels in Cushing, Oklahoma, where the WTI is priced.

• Increasing oil supplies were coincided with inadequate transportation infrastructure, inability to export crude oil and unsuitable refinery configurations.

ConclusionsAdded Value:

• The evolution of the U.S. oil transportation system trying to adapt to the new standards, came as a shock to the oil market and the oil price spread experienced a structural change on 13 March 2014.

• The spread formed a new stationary relationship with WTI trading at a slight discount compared to Brent.

• There is a positive relationship between the price spread and the oil pipeline flows from PADD2 to PADD3. Expansions and reversals of the pipeline system are associated with an upward pressure to the price spread.

• There is an inverse relationship between the price spread and the U.S. oil production as well as the Canadian oil imports. Increasing oil supply causes the price spread to decrease.

Thank you for your attention!