The interdependence of financial markets - econometric modeling and estimation

This thesis analyses symmetric and asymmetric interdependencies of financial markets through time for returns and variances. The work contributes to the literature in several areas: Symmetric interdependencies (correlations) are examined in the first part where a new bivariate correlation estimator is introduced. This estimator is shown to be more flexible than existing multivariate GARCH models and thus not prone to potential misspecifications. Furthermore, merits and shortcomings of existing multivariate GARCH models are discussed and evaluated in a simulation study. Differences of daily and monthly returns regarding the persistence and the asymmetry of correlations are additionally analysed. The second part studies asymmetric interdependencies (spillovers) and proposes a classification. An adequate estimation framework to investigate spillovers conditional on time and conditional on the values of the dependent variable is presented and it is shown under which conditions spurious interdependencies can occur. Finally, the existence of contagion in financial markets is examined.