Asymmetry and nonlinearity in forecasting multivariate stock market volatility

This cumulative dissertation studies various approaches to improve stock market volatility forecasts based on nonlinearity and asymmetric dependence modeling as well as new innovative data sources. Studying multivariate dependence patterns using a vine copula approach and incorporating Google search data as measure of investor attention in a framework of empirical similarity significantly improves volatility forecasts based on different statistical and economic measures. The importance of accurate volatility forecasts in portfolio- and risk management is highlighted in several economic applications and empirical studies.