Elsevier

Energy Economics

Volume 32, Issue 6, November 2010, Pages 1445-1455
Energy Economics

Analyzing and forecasting volatility spillovers, asymmetries and hedging in major oil markets

https://doi.org/10.1016/j.eneco.2010.04.014Get rights and content

Abstract

Crude oil price volatility has been analyzed extensively for organized spot, forward and futures markets for well over a decade, and is crucial for forecasting volatility and Value-at-Risk (VaR). There are four major benchmarks in the international oil market, namely West Texas Intermediate (USA), Brent (North Sea), Dubai/Oman (Middle East), and Tapis (Asia-Pacific), which are likely to be highly correlated. This paper analyses the volatility spillover and asymmetric effects across and within the four markets, using three multivariate GARCH models, namely the constant conditional correlation (CCC), vector ARMA–GARCH (VARMA–GARCH) and vector ARMA–asymmetric GARCH (VARMA–AGARCH) models. A rolling window approach is used to forecast the 1-day ahead conditional correlations. The paper presents evidence of volatility spillovers and asymmetric effects on the conditional variances for most pairs of series. In addition, the forecast conditional correlations between pairs of crude oil returns have both positive and negative trends. Moreover, the optimal hedge ratios and optimal portfolio weights of crude oil across different assets and market portfolios are evaluated in order to provide important policy implications for risk management in crude oil markets.

Introduction

Over the past 20–30 years, oil has become the biggest traded commodity in the world. In the crude oil market, oil is sold under a variety of contract arrangements and in spot transactions, and is also traded in futures markets which set the spot, forward and futures prices. Crude oil is usually sold close to the point of production, and is transferred as the oil flows from the loading terminal to the ship FOB (free on board). Thus, spot prices are quoted for immediate delivery of crude oil as FOB prices. Forward prices are the agreed upon price of crude oil in forward contracts. Futures price are prices quoted for delivering in a specified quantity of crude oil at a specified time and place in the future in a particular trading centre.

The four major benchmarks in the world of international trading today are: 1) West Texas Intermediate (WTI), the reference crude for USA, (2) Brent, the reference crude oil for the North Sea, (3) Dubai, the benchmark crude oil for the Middle East and Far East, and (4) Tapis, the benchmark crude oil for the Asia-Pacific region. Volatility (or risk) is important in finance and is typically unobservable, and volatility spillovers appear to be widespread in financial markets (Milunovich and Thorp, 2006), including energy futures markets (Lin and Tamvakis, 2001). These results hold even when markets do not necessarily trade at the same time. Consequently, a volatility spillover occurs when changes in volatility in one market produce a lagged impact on volatility in other markets, over and above local effects. Volatility spillovers and asymmetries among those four major benchmarks are likely to be important for constructing hedge ratios and optimal portfolios. As research has typically focused on oil spot and futures prices to the neglect of forward prices, this paper analyses all three oil prices.

Accurate modelling of volatility is crucial in finance and for commodity. Shocks to returns can be divided into predictable and unpredictable components. The most frequently analyzed predictable component in shocks to returns is the volatility in the time-varying conditional variance. The success of the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model of Engle, 1982, Bollerslev, 1986 has subsequently led to a family of univariate and multivariate GARCH models which can capture different behavior in financial returns, including time-varying volatility, persistence and clustering of volatility, and the asymmetric effects of positive and negative shocks of equal magnitude. In modelling multivariate returns, such as spot, forward and futures returns, shocks to returns not only have dynamic interdependence in risks, but also in the conditional correlations which are key elements in portfolio construction and the testing of unbiasedness and the efficient market hypothesis. The hypothesis of efficient markets is essential for understanding optimal decision making, especially for hedging and speculation.

Substantial research has been conducted on spillover effects in energy futures markets. Lin and Tamvakis (2001) investigated volatility spillover effects between New York Mercantile Exchange (NYMEX) and International Petroleum Exchange (IPE) crude oil contracts in both non-overlapping and simultaneous trading hours. They found that substantial spillover effects exist when both markets are trading simultaneously, although IPE morning prices seem to be affected considerably by the close of the previous day on NYMEX. Ewing et al. (2002) examined the transmission of volatility between the oil and natural gas markets using daily returns data, and found that changes in volatility in one market may have spillovers to the other market. Sola et al. (2002) analyzed volatility links between different markets based on a bivariate Markov switching model, and discovered that it enables identification of the probabilistic structure, timing and the duration of the volatility transmission mechanism from one country to another.

Hammoudeh et al. (2003) examined the time series properties of daily spot and futures prices for three petroleum types traded at five commodity centres within and outside the USA by using multivariate vector error-correction models, causality models and GARCH models. They found that WTI crude oil NYMEX 1-month futures prices involves causality and volatility spillovers, NYMEX gasoline has bi-directional causality relationships among all the gasoline spot and futures prices, spot prices produce the greatest spillovers, and NYMEX heating oil for 1- and 3-month futures are particularly strong and significant. Chang et al. (2009) examined multivariate conditional volatility and conditional correlation models of spot, forward, and futures returns from three crude oil markets, namely Brent, WTI and Dubai, and provided evidence of significant volatility spillovers and asymmetric effects in the conditional volatilities across returns for each market.

Of the four major crude oil markets, only the most well known oil markets, namely WTI and Brent, the light sweet grade category, have spot, forward and futures prices, while the Dubai and Tapis markets, the heavier and less sweet grade category, have only spot and forward prices. It would seem that no research has yet tested the spillover effects for each of the spot, forward and futures crude oil prices in and across all markets, or estimated the optimal portfolio weights and optimal hedge ratios for purposes of risk diversification.

Spot, futures and forward oil markets have different fundamentals and contract tradability and liquidity, and thus different volatility behavior. Forward markets are usually less volatile than spot and futures markets. It would therefore be interesting to determine if this stylized characteristic holds across the major oil benchmarks.

Several multivariate GARCH models specify risk for one asset as depending dynamically on its own past and on the past of other assets (see McAleer, 2005). da Veiga et al. (2008) analyzed the multivariate vector ARMA–GARCH (VARMA–GARCH) model of Ling and McAleer (2003) and vector ARMA–asymmetric GARCH (VARMA–AGARCH) model of McAleer et al. (2009), and found that they were superior to the GARCH model of Bollerslev (1986) and the GJR model of Glosten et al. (1992).

This paper has two main objectives, as follows: (1) We investigate the importance of volatility spillovers and asymmetric effects of negative and positive shocks of equal magnitude on the conditional variance for modelling crude oil volatility in the returns of spot, forward and futures prices within and across the Brent, WTI, Dubai and Tapis markets, using multivariate conditional volatility models. The spillover effects between returns in and across markets are also estimated. A rolling window is used to forecast 1-day ahead conditional correlations, and to explain the conditional correlations movements, which are important for portfolio construction and hedging. (2) We apply the estimated results to compute the optimal hedge ratios and optimal portfolio weights of the crude oil portfolio, which provides important policy implications for risk management in crude oil markets.

The plan of the paper is as follows. Section 2 discusses the univariate and multivariate GARCH models to be estimated. Section 3 explains the data, descriptive statistics and unit root tests. Section 4 describes the empirical estimates and some diagnostic tests of the univariate and multivariate models, and forecasts of 1-day ahead conditional correlations. Section 5 presents the economic implications for optimal hedge ratios and optimal portfolio weights. Section 6 provides some concluding remarks.

Section snippets

Econometric models

This section presents the constant conditional correlation (CCC) model of Bollerslev (1990), the VARMA–GARCH model of Ling and McAleer (2003) and VARMA–AGARCH model of McAleer et al. (2009). These models assume constant conditional correlations, and do not suffer from the problem of dimensionality, as compared with the VECH and BEKK models, and also possess regularity and statistical properties, unlike the DCC model (see McAleer et al., 2008, Caporin and McAleer, 2009, Caporin and McAleer, 2010

Data

The univariate and multivariate GARCH models are estimated using 3009 observations of daily data on crude oil spot, forward and futures prices in the Brent, WTI, Dubai and Tapis markets for the period 30 April 1997 to 10 November 2008. All prices are expressed in US dollars. In the WTI market, prices are crude oil-WTI spot cushing price ($/BBL), crude oil-WTI one-month forward price ($/BBL), and NYMEX one-month futures prices. The prices in the Brent market are crude oil-Brent spot price FOB

Empirical results

From Table 3, Table 4, the univariate ARMA(1,1)–GARCH(1,1) and ARMA (1,1)–GJR(1,1) models are estimated to check whether the conditional variance follows the GARCH process. In Table 3, not all the coefficients in mean equations of ARMA(1,1)–GARCH(1,1) are significant, whereas all the coefficients in the conditional variance equation are statistically significant. Table 4 shows that the long-run coefficients are all statistically significant in the variance equation, but rbrefu (Brent futures

Implications for portfolio design and hedging strategies

This section presents optimal hedge ratios and optimal portfolio weights among crude oil returns and across markets. Theoretically, hedging involves the determination of the optimal hedge ratio. One of the most widely used hedging strategies is based on the minimization of the variance of the portfolio, the so-called minimum variance hedge ratio (see, for example, Kroner and Sultan, 1993, Lien and Tse, 2002, Chen et al., 2003). In order to minimize risk, the dynamic hedge ratio, based on

Conclusion

The empirical analysis in the paper examined the spillover effects in the returns on spot, forward and futures prices of four major benchmarks in the international oil market, namely West Texas Intermediate (USA), Brent (North Sea), Dubai/Oman (Middle East) and Tapis (Asia-Pacific), for the period 30 April 1997 to 10 November 2008. Alternative multivariate conditional volatility models were used, namely the CCC model of Bollerslev (1990), VARMA–GARCH model of Ling and McAleer (2003), and

Acknowledgements

The authors wish to thank two referees for helpful comments and suggestions, and Felix Chan and Abdul Hakim for providing the computer programs. For financial support, the first author is most grateful to the National Science Council, Taiwan, the second author wishes to thank the Australian Research Council, National Science Council, Taiwan, and a Visiting Erskine Fellowship at the University of Canterbury, New Zealand, and the third author acknowledges the Faculty of Economics, Maejo

References (26)

  • T. Bollerslev

    Modelling the coherence in short-run nominal exchange rate: a multivariate generalized ARCH approach

    Review of Economics and Statistics

    (1990)
  • T. Bollerslev et al.

    Quasi-maximum likelihood estimation and inference in dynamic models with time-varying covariances

    Econometric Reviews

    (1992)
  • Caporin, M. and M. McAleer, 2009, Do we really need both BEKK and DCC? A tale of two covariance models. Available at...
  • Cited by (124)

    • From black gold to financial fallout: Analyzing extreme risk spillovers in oil-exporting nations

      2024, Journal of International Financial Markets, Institutions and Money
    View all citing articles on Scopus
    View full text