Oil price shocks, equity markets, and contagion effect in OECD countries.

AuthorGuesmi, Khaled
  1. Introduction

    Since 2002, the commodity market has experienced 'financialization', as financial investors represent more than 80% of the total investors (Gao and Suss, 2015). The most strategic and most volatile commodity is crude oil, leading to increased attention being paid to its prices with respect to those of other commodities. Interestingly, the attention to the dynamics of oil prices has been growing since the end of the 1990s, when different financial crises and events occurred, leading to boom or bust in international trade and, consequently, proving the high volatility of oil prices.

    The theory of equity valuation might explain the impact of oil price fluctuations on stock prices. According to this theory, stock prices are obtained by discounting all expected future cash flows at the investors' required rate of return. For instance, a negative oil shock may reduce the corporate cash flow and rate of return. In addition, stock prices could be affected by oil price fluctuations through the channels of corporate earnings, low stock prices, and economic conditions. Based on these channels, crude oil presents some specificities compared to other commodities, especially those used in production processes.

    Numerous studies have analysed the specificities of the relationship between oil and financial markets, finding a negative relationship between oil price and stock market returns. For instance, Arouri et al. (2011) use a multivariate GARCH to investigate volatility spillovers between oil and different stock market sectors in the US and Europe, using a weekly dataset from January 1998 to December 2009. They find that oil prices fluctuations affect both the European and the US stock markets. More recently, Chang et al. (2013) also use a multivariate GARCH model to study the volatility spillovers effect between oil prices fluctuations and the US and UK stock markets. Their results show no significant evidence of volatility spillovers between oil prices and stock markets. Further, Creti et al. (2013) analyse the time-varying correlations between oil prices, commodities, and stock market indexes, concluding that commodity price correlations increase after 2003, limiting hedging substitutability in portfolios, and become more significant after the financial crisis period in 2008. Using a Wavelet approach, Roboredo and Rivera-Castro (2014) study the daily connection between oil prices and the US and European financial markets. Their main finding supports that the oil price changes did not have a substantial effect on the stock market returns in the precrisis period. Sadorsky (2014) uses VARMA-AGARCH and DCCA-GARCH specifications to model the volatilities and conditional correlation dynamics between emerging stock markets, and copper, oil, and wheat prices. The results show the correlations between assets increased considerably after the financial crises and the hedge ratios vary extensively over the sample period, showing that hedged positions should be efficient frequently. More recently, Zhang (2017) investigates the oil-stock relationship from a global perspective. Based on the Diebold and Yilmaz (2009, 2012, 2014) measure of the oil-stock connection for six markets, they show that overall, oil price shocks have a limited effect on the world financial system. Huang et al. (2018) investigate the co-movement between oil stocks based on a frequency approach, from a multivariate perspective. Through different oil prices (Brent, Dubai, Minas, and OPEC, Shanghai Composite index), the authors show the relationship between oil stocks is tremendously different in the short run. Interestingly, the results support that investment in oil prices on the Brent, OPEC, and Chinese stock markets might be sources of risk reduction.

    Some other studies analyse this issue by distinguishing between oil importing and exporting countries. For example, Filis et al. (2011) study co-movements and time varying correlations between oil and stock markets for both oil importing and exporting countries using a multivariate GARCH model. They conclude that conditional variances between stock and oil prices do not vary significantly between oil exporting and importing countries. Creti et al. (2014) use the frequency approach to study the time-varying correlations between oil prices and stock markets also by distinguishing between oil exporting and importing countries. Their results show the interdependence between oil prices and stock markets is more pronounced for oil exporters. Guesmi and Fattoum (2014), based on multivariate asymmetric GARCH models, show an important impact of oil prices on both oil importing and exporting countries.

    Our study extends these previous studies by proposing a more flexible framework to measure the volatility spillover effect between oil and stock markets. Specifically, we employ the multivariate c-DCC-FIAPARCH specification to consider main financial volatility features. Our specification allows more flexibility for the conditional variance process, as it reacts asymmetrically to positive and negative shocks. Moreover, our approach captures the long-range volatility dependence. The convenience of our empirical framework is investigated through considering the dynamic interactions between the Brent oil market and 17 OECD stock markets over the period March 16, 1998-February 23, 2018, which is characterized by several peaks and troughs of oil prices, as well as several periods of financial turmoil.

    Our analysis reveals two main findings about the interactions between oil prices and the major OECD stock markets. The impact of oil shocks on stock markets is more pronounced during periods of global turmoil. Our analysis contributes to the literature by identifying two types of correlation coefficient signs between stock and oil markets. First, we consider the US terrorist attack (2001) as the main source of the negative correlation between oil and stock markets. The positive trend is identified during other periods coinciding with aggregate demand-side oil price shocks, such as the Asian crisis (1997-1998), Chinese economic growth (2002), and the global financial crisis (2007-2008). Our empirical method allows us to show the repercussions of these phenomena are not symmetric for the entire sample. Diversification opportunities were thus generally decreasing in all the studied countries. Moreover, based on the shape of the dynamic conditional correlation, we characterize five groups of countries for which the dynamic correlations between oil and stock markets are similar. (1) Therefore, although there exists a form of spillover effect among all markets, the diffusion of shocks and volatility does not reveal contagion, as the evolution of the dynamic correlations between oil and stock markets remain segmented geographically. Consequently, we do not observe a proper 'contagion effect', as least as the phenomenon defined by Forbes and Rigobon (2002), that is, the co-movement increase between markets after crises.

    The rest of the paper is organized as follows. Section 2 introduces the empirical method used to assess the links between oil and stock markets. Section 3 describes the data and discusses the empirical results. Section 4 provides concluding remarks.

  2. Methodology

    Let [x.sub.t] be an (18x18) vector composed from 17 OECD countries' stock markets prices and the oil price containing the return series in a conditional mean equation as follows:

    [x.sub.t] = [[micro].sub.t] + [[epsilon].sub.t], (1)

    where [[micro].sub.t] = E[[x.sub.t][parallel][[PI].sub.t-1] is the conditional expectation of xt based on previous

    information [[PI].sub.t-1] * [[epsilon].sub.t] is the error vector, assumed to be conditional multivariate normally distributed. [[epsilon].sub.t] has a zero mean and variance-covariance matrix [H.sub.t] [equaivalent] {[h.sub.t]}.

    The model can be estimated through maximum likelihood methods if [H.sub.t] is positive definite for all [[epsilon].sub.t] values in the sample. Furthermore, we assume [[micro].sub.t] is linearly specified, as follows:

    [[micro].sub.t] = [[PHI].sub.0] + [[PHI].sub.1][x.sub.t-i], [for all]i, (2)

    where [[PHI].sub.0] is a constant and [[PHI].sub.1] measures the ARCH effect in the data series.

    In this paper, we suggest using the DCC-GARCH process, as defined by Engle (2002), to model the dynamic conditional correlation between stock and oil markets. This model follows the generalized fractional cointegration process and the conditional variance-covariance matrix, which can be written as:

    [H.sub.t] = [D.sub.t][R.sub.t][D'.sub.t], (3)

    where [H.sub.0] is the (18x18) symmetric matrix of dynamic conditional correlations and [D.sub.t] a diagonal matrix of time-varying standard deviation from univariate GARCH models. These matrices can be written as:

    [mathematical expression not reproducible] (4)

    where [Q.sub.t] = ([q.sub.ijt]) is a symmetric positive matrix, which is assumed to vary according to a GARCH-type process, with [bar.Q] being an (18x18) unconditional variance matrix of standardized residuals [[eta].sub.i,t]. [[theta].sub.1] and [[theta].sub.2] capture the effects of shocks to dynamic correlations. The correlation coefficient is defined as follows:

    [[rho].sub.ij,t] = [q.sub.ij,t]/[square root of [q.sub.ij,t] [q.sub.jj,t]. (5)

    To develop the lack of regularity and potential bias for the estimated parameters of the DCC-GARCH model of Engle (2002), Aielli (2008) suggests a corrected dynamic conditional correlation (c-DCC) process. The conditional standard deviation of the FIAPARCH process is expressed as:

    [mathematical expression not reproducible] (6)

    where

    * [h.sub.i,t] refers to the conditional variance of [x.sub.i,t]; [[omega].sub.i]. is the mean of the process;

    * [mathematical expression not reproducible] represents the fractional degree of integration of [h.sub.i,t];

    * [[psi].sub.i] (L) and [[phi].sub.i] (L) are the lag polynomials of the respective orders P and K;

    *...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT