Modelling oil price volatility before, during and after the global financial crisis

• DOI: http://doi.org/10.1111/opec.12037
• Date: 01 December 2014
• Author: Afees A. Salisu
• Published date: 01 December 2014

Modelling oil price volatility before, during

and after the global financial crisis

Afees A. Salisu, PhD

Department of Economics & Centre for Econometric & Allied Research (CEAR), University of Ibadan,

Ibadan, Oyo State +234 Nigeria. Email: aa.salisu@mail.ui.edu.ng, aa.salisu@cear.org.ng.Tel.:

+2348034711769

Abstract

In this paper, weevaluate the comparative performance of volatility models for oil price using daily

returns of crude oil price. The innovationsof this paper are in three folds: (i) we consider two promi-

nent oil prices namely Brent and West Texas Intermediate (WTI); (ii) weanalyse these prices across

three subsamples namely periods before, during and after the global financial crisis; and (iii) wealso

analyse the comparativeperfor mance of both symmetric and asymmetric volatilitymodels for these

oil prices. We find inconsistent patterns in the performance of the volatility models over the

subsamples. On the average, however, we find evidence of leverage effects in both oil prices and

therefore, investors in the oil market react to news. Specifically, we find that bad news in the oil

market increased volatilityin cr ude oil price than good news.We also find high levelof persistence in

the volatility of WTI and Brent although the latter appears more persistent than the former while the

period of global financial crisis recorded the highest level of persistence in both prices.Also, we find

that during the global financial crisis, risk averse investorsshifted assets from the oil market to other

less risky assets.

1. Introduction

The concept of volatility in oil price is increasingly gaining prominence both in theory and

practice.The reasons for this development are obvious; oil price data are available at a high

frequency and therefore, there is increasing evidence of the presence of statistically sig-

nificant correlations between observations that are large distance apart; and also in con-

nection with the high frequency of oil price data, there is possibility of time varying

volatility (referred to as conditional Heteroscedasticity) (see Harris and Sollis, 2005).

More practically, variability in the oil price implies huge losses or gains to oil-producing

and oil-exporting nations particularly the oil-dependent economies and hence are con-

fronted with economic instability and huge losses or gains to independent investors in

the oil markets and hence they are confronted with greater uncertainty. Thus, both the

JEL Classification: C22, G01, Q40

469

© 2014 Organization of the Petroleum Exporting Countries. Published by John Wiley & Sons Ltd, 9600 Garsington

Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

government and profit-maximising investorsare keenly interested in the extent of volatil-

ity in oil price to make policy/investment decisions.Therefore, a measure of volatility in

oil price provides useful information both to the investors in terms of howto make invest-

ment decisions and relevant authorities in terms of how to formulate appropriate policies.

A more serious concern, however, centres on how to model oil price when confronted with

such volatility.

Evidently,the modelling and forecasting of oil price volatility have followed different

dimensions in the literature (see Sadorsky, 2006 and Narayan and Narayan, 2007 for a

survey of the literature).1Generalising the model of oil price volatility, notwithstanding

significant peculiarities, may lead to misleading inferences. Essentially, in the present

study, wemodel oil price volatility before, during and after the global financial crisis.

Narayan and Narayan’s (2007) paper appears to be the only notable paper that has

attempted to model and forecast oil price volatility using various subsamples in order to

judge the robustness of their results; however, there was no justification for the consid-

eration of such subsamples. In the present study, our choice of subsamples wasmotivated

by the incidence of the global financial crisis and the intention is to ascertain whether

the incidence of this crisis altered the modelling framework for dealing with oil price

volatility.

In addition, many research worksusually consider one type of price when dealing with

oil price volatility (see for a survey of the literature, Sharma, 1998; Sadorsky, 2006; and

Narayan and Narayan, 2007). In the present study, we consider two prominent oil prices

namely Brent andWest Texas Intermediate (WTI thereafter) in order to evaluatethe behav-

iour of these prices over the subsamples and to also check robustness of our results. We

also analyse the comparative performance of both symmetric and asymmetric volatility

models for the two oil prices under each subsample.

In this study, our analyses are carried out in three phases. The first phase deals with

some pretests to ascertain the existence of volatility in oil price. TheAutoregressive Con-

ditional Heteroscedasticity (ARCH) Lagrangian Multiplier (LM) test proposed by Engle

(1982) coupled with some descriptive statistics are employed in this regard.The second

phase proceeds to estimation of both symmetric and asymmetric volatility models. Model

selection criteria such as Schwartz Information Criterion (SIC), Akaike Information

Criterion (AIC) and Hannan–Quinn Information Criterion (HQC) are used to determine

the model with the best fit. The third phase provides some post-estimation analysesusing

the same ARCH LM test to validate the selected volatilitymodels.

We find inconsistent patterns in the performance of the volatility models over the

subsamples. On the average, however, we find evidence of leverage effects on both oil

prices and therefore the asymmetric models appear superior to the symmetric models.This

suggests that investorsin the oil market react to news. Specifically, wefind that bad news in

the oil market increased volatility in crude oil price than good news.We also find high level

Afees A. Salisu470

OPEC Energy Review December 2014 © 2014 Organization of the Petroleum Exporting Countries