The analysis of return and volatility spillovers and integration between Turkish stock market and some developed and developing stock markets (2009-2014)

In this study, the return and volatility spillover and dynamic linkages between Turkish stock market and major developed and developing stock markets including USA, England, Brazil, Mexico, India, Japan, Hong Kong and Russia is investigated by using Johansen cointegration test, VAR-EGARCH model and VAR analysis methods. For this study the most recent data of daily closing stock prices indices taken for the period 01.01.2009 to 30.06.2014. This study has used Augmented Dickey-Fuller (ADF) tests for judging the stationarity of the data series. ADF test reveals that all the variables under study are non-stationary at level, but integrated of order 1. After the Augmented Dickey-Fuller test, the long-run cointegration relationship between Turkish stock market and other selected stock markets is investigated by using Johansen cointegration test. According to the results obtained; although there is cointegration relation between Turkey and some of other stock exchanges under study; England, Hong Kong, India and Mexico; a meaningful cointegration relation does not exist between Turkey and USA, Brazil, Japan and Russia for the period between 2009 and 2014. Generally, developed markets (U.S., Japan and European markets) are used as exogenous (control) variable in this type of studies. However in current scenario of economic trade and financial liberalization it is plausible that the emerging markets may also affect the other countries' stock markets. The spillover from one market to another market may not be limited to the first movements of the return series. It is also possible that the volatility (the second movement) in one market gets transmitted to another market. Therefore, we include the volatility spillover as well as return spillover in this study and examine the magnitude of return and volatility spillovers from selected developed and developing stock markets to Turkish stock market and whether these spillover effects display asymmetric characteristics through bivariate EGARCH models. For this purpose, first we model stock market returns using an AR(p)-EGARCH(p,q). The standardized residuals are obtained from this model and their squared values are used as volatility shocks from the stock markets. The lagged values of these shocks are included in the volatility specification of Turkish stock market returns to allow volatility spillover from the other selected markets to Turkish stock market, while the lagged returns of these stock markets are included in the mean equation of turkish stock market returns to allow the return-spillover. Suitable dummies are used representing negative shocks to allow sign asymmetries. We construct a benchmark model, which includes both return and volatility spillover from the other selected markets to Turkish stock market, but does not include any asymmetry in spillover effect. The results from this model are compared with those of other models with asymmetric spillover to examine whether the asymmetric spillover gives a better fit to the data. The evidence is found that Turkish stock market is affected from developed and developing stock markets under study in terms of both return and volatility. The partial effects of the stock markets of USA and Brazil in the volatility of Turkish stock market and the partial effects of the stock markets of USA, England and Russia in the return of Turkish stock market are higher than other markets under study. Furthermore, the study observes a significant asymmetry in this spillover effects. Returns and volatility in Turkish stock market are generally more sensitive to negative shocks in the other selected stock markets rather than the positive shocks. Positive shocks in the stock markets of England, Russia and Hong Kong do not affect the volatility in the Turkish stock market, but negative shocks significantly increase the volatility. Financial liberalization policies, increasing international investment, developments in communication and computer technologies almost removed geographical and physical barriers that separate stock exchange markets. Today, a market system that operates continuously for 24 hours and that does not have geographical boundaries has emerged. A negative development that appears in a market in the global system can be reflected upon other markets. Therefore, in the second part of this study, long-term and short-term relations between the Turkish stock market and the stock markets of other selected countries are analyzed through a correlation analysis and the nine-variable VAR (Auto-regressive) model. The results obtained from the VAR model were interpreted by the Granger Causality Test, Impulse-Response Analysis and the Variance Decomposition methods for the purpose of explaining relations between the Turkish stock market and other countries' stock markets. Correlation matrix was applied for finding associationship between Turkish stock market and other selected stock markets which shows evidence of integration of Turkish and USA, India, Hong Kong, England and Mexico stock markets. After the correlation analysis, the Johansen cointegration approach was applied for checking the long run relationship between all variables which shows no evidence of cointegration among these stock markets but short-run causality could not be rejected. After the Johansen co-integration test, firstly the dynamic interactions between the returns of stock markets have been investigated in this study by using VAR methods. Granger Causality test reveals that Turkish stock market's return is granger caused by USA, Brazil and Russia stock markets' returns. Variance decomposition provides the proportion of variance in the dependent variables that is due to their own shocks versus shocks from other variables. Our variance decomposition analysis implies that variances in stock market's return of Turkey are due to their own market innovation and USA and England of other selected stock markets have significant contribution to variances in stock market's return of Turkey. According to variance decomposition results, Turkish stock market explains 66.08% of variance of its return itself and 16.74%, 7.90%, 3.09% and 2.63% variances of Turkish stock market's return are being explained by stock markets of USA, England, Russia and Brazil respectively. Furthermore, we use impulse response functions to analyze the relative impact of shocks in the other selected stock markets on Turkish stock market's return. Impulse response represents the responsiveness of the dependent variable in the VAR to shocks from each of the variables. Granger causality and variance decomposition does not give any information about the sign of relationship or how long it requires for these effects to takeplace. Impulse response answers these questions. Evidence suggests that Turkish stock market's return responses significantly to schocks in the stock markets of USA and England. After the investigation of relations between the returns of Turkish stock market and other countries' stock markets under study, we have investigated the potential dynamic interactions among our variables' volatilities by using VAR model and we have seen that volatility index of stock markets that employed in this study are found to Granger cause Turkish stock market volatility index except stock markets of Hong Kong and England. Also we have found that a shock to the volatility index of Brazil, USA, England, Russia stock markets have significant effect on the Turkish stock market volatility by using impulse-response analysis. On the basis of the variance decomposition analysis we have found that a substantial fraction of variance in the volatility of Turkish stock market is explained by past changes in the volatility of stock market of Turkey, Brazil and USA. According to variance decomposition results, Turkish stock market explains 70% of variance of its volatility itself and 13.6%, 6.5% variances of Turkish stock market's volatility are being explained by stock markets of Brazil, USA respectively.