TopIntroduction
A great deal of research has focused upon the interdependence among international financial markets; particularly; in terms of how shocks or variations are transmitted across global stock markets (Narayan & Smyth, 2004; Fraser & Oyefeso, 2005; Cappiello, Engle, & Sheppard, 2006; Chen & Shen, 2007; Yu & Hassan, 2008; Ahlgren & Antell, 2010; Qiao, Li, & Wong, 2011) based on the assumption that the simultaneous dynamics of asset returns are well captured by a linear vector-autoregression (VAR) or a vector error-correction model (VECM) of Johansen (1988, 1991). The obtained empirical results indicated that the world stock markets have become more closely linked in recent years due to the development of global cooperation, financial market liberalization and the advances in information processing technology (Qiao, Li, & Wong, 2011). The VAR and VECM (Johansen, 1988, 1991) are popular statistical methods since they can systematically describe the dynamics among several variables, and were widely employed for multivariable systems modeling and forecasting in science (Chandra & Al-Deek, 2009; García-Ascanio & Maté, 2010; Wong & Ng, 2010; Xu & Moon, 2013) and economics (Jore, Mitchell, & Vahey, 2010; Clark, 2011; Das, Gupta & Kabundi, 2011; Korobilis, 2013).
The advantage of using VAR and VECM as statistical models is to account for the linear dynamics of the temporal causal relationship between stock markets that compose a system (Masih & Masih, 1999; Laopodis, 2011). In particular, they allow for examining both the short and long term dynamic causal linkages among variables of the system (Masih & Masih, 1999). Moreover, they are useful to capture long-run information often ignored in systems (Masih & Masih, 1999; Laopodis, 2011). For instance, VECM searches evidence of cointegration between variables in the system to find if they share common stochastic trend at long-run, or share deviations at short-run, or share both (Laopodis, 2011). The importance of a statistically significant long-run relationship among variables in the system lies in the existence of common stochastic trend, and; therefore; they tend to revert to such relationship after some short-run fluctuations (Laopodis, 2011). As a result, VECM provide long-run system parameters useful to ascertain the fundamental information content of the variables in the system (Tswei, 2013). In addition, if the short-run co-movements exist among stock markets that form the system under study; then, equity managers may design their portfolios to try to win the market (Chu, 2011). All these types of information regarding the relationships between stock markets that form the system could be used to predict each stock market future value.