Uzun hafıza ve bulaşma etkisinin yapısal kırılmaların varlığıyla analiz edilmesi

This thesis presents two essays on long memory and shift-contagion effect. The first chapter investigates long memory properties for Germany, France, Turkey, the Netherlands, England, Italy, and US stock market returns within the framework of weak-form efficient market hypothesis and asset allocation. Weekly data are used for the period January 1998- December 2019 in both chapters of the thesis. The study uses the ARFIMA-FIEGARCH model to examine the long memory in seven stock indices returns. In addition, change points are detected by employing of Bai-Perron structural test. The findings suggest that the stock returns are highly persistent in almost all European and US stock indices except for Turkey. These markets are interdependent which violates the weak efficiency of the market hypothesis. This would mean that historical returns can contribute to predicting future returns in those markets. Then, the study compares the persistence of volatility with and without structural breaks. Due to its structural breaks, the results are mixed. The finding of the study shows that structural break effectively reduces the persistence of volatility in the countries of Germany, France, England, and the Netherlands. Whereas the long memory value for each index of FTSEMIB, BIST100, and S&P500 does not reduce the persistence of volatility or long memory. The findings offer valuable insights and create an opportunity for skilled investors to make speculative profits consistently. The second chapter examines the shift-contagion effect of stock markets with the presence of structural breaks for asset allocation and risk management. I propose the bivariate DCC-GARCH model to test shift contagion between European countries of Germany, France, Italy, Netherlands, England, and Turkey indices with the US index. Our results suggest that the dynamic conditional correlations do not show a positive increase between European and US markets as well as a weak in market co-movements. The weak values of correlations suggest that they enable investors to get long-term benefits by diversifying over across markets. The relationship of the US stock market with each European market is not contagious but the markets are only interdependence. We conclude that investors would achieve a greater advantage of diversification in European and US stock markets. However, the study found shift contagion between US and Turkey stock markets co-movement. So, a higher degree of relationship between markets would reduce the diversification benefits in the period of crisis. Keywords: DCC-GARCH Model, FIEGARCH, Contagion Effect, Long memory