Stock Exchange Volatility Transmissions between Turkey and the Major Financial Centers

Abstract

In the last thirty years, volatility modeling in financial time series has drawn considerable attention in the literature of financial econometrics. The workhorse model of volatility has been the ARCH model and its generalized counterparts in that period. Although these models have proven to be useful in capturing some stylized facts of volatility in financial time series such as the cluster effect and have been applied to diverse areas with success such as modeling systematic risk changes in the market, asset pricing via models like I-CAPM or APT, or testing market efficiency; they have a major drawback. They are univariate. That is why, a typical ARCH or GARCH-type model allows to detect the effects of prior shocks and conditional variances in the errors of the same variable. However, thanks to a growing literature, it is now known that volatility spillovers (both in the form of shock and conditional variance transmissions) between two or more variables might exist. In this paper, I study the potential volatility interactions between the major stock exchange indices in the global financial centers and in Turkey. To be precise, the dataset includes observations on the returns of the leading indices in Turkey, UK, Germany, Japan and the US. For that purpose, I employ a multivariate model called the Extended Constant Conditional Correlation GARCH (ECCC-GARCH) model whose identification properties are developed by Nakatani and Terasvirta (N-T). ECCC-GARCH model is a useful tool because N-T have developed a powerful LR test for this model which allows to check whether there really exist volatility spillovers between the variables in a dataset or not. Results of this study indicate the existence of universal shock transmissions from the stock exchange returns in the global financial centers to those in Turkey. There also exist conditional variance interactions between the USA, UK and Turkey. These results suggest that the stock exchange in Turkey well integrated to the world but not quite suitable to be used by stock exchange investors in the USA and UK for the purpose of global risk diversification.

Keywords: Multivariate GARCH models, Volatility Spillovers, Conditional Variance

JEL Classification: C58, C32