The crises of recent years (sub-prime crises, financial-/banking crises, European government crises) illustrated the limits of traditional portfolio models. Based on the assumption of independent and identically distributed returns these models neglected the dynamics and risks of significant share price losses in falling markets. Consequently, the portfolio which should be generated by these models suffered heavy losses and could not fulfill the requirements of risk diversification.
The aim of this thesis is to present a class of models (markov switching models) for modelling asset returns without constant parameters, but capturing the asymmetric characteristics of asset classes in rising and falling markets. Furthermore, it confirms that these models are able to depict other stylized facts like skewness, excesscurtosis, volatility clustering as well as leverage effects. These are so-called Markov switching models enabling the modelling of market situations by using the finite Markov chain. Moreover, the asset returns show a different distribution that varies depending on the current market situation.
As selected example of this model class, the so called Basic Markov switching model with conditionally independent normally distributed returns and a two-state Markov chain is being analyzed and applied in a portfolio optimization for the structuring of asset classes.
The results show that using such a model yields lower portfolio risk compared to standard approaches. Furthermore, an explicit consideration of varying state probability in the optimization framework yields in addition better portfolio performance.
By using this process it will be also possible to analyse in which way the selected model can be adjusted to investor specific assumptions about the probability distribution of market situations as well as to assumptions about distribution parameters of return distribution.
«
The crises of recent years (sub-prime crises, financial-/banking crises, European government crises) illustrated the limits of traditional portfolio models. Based on the assumption of independent and identically distributed returns these models neglected the dynamics and risks of significant share price losses in falling markets. Consequently, the portfolio which should be generated by these models suffered heavy losses and could not fulfill the requirements of risk diversification.
The aim of...
»