Not logged in.
Quick Search - Contribution
Contribution Details
| Type | Master's Thesis |
| Scope | Discipline-based scholarship |
| Title | International Stock Portfolios and the Role of Foreign Exchange Risks |
| Organization Unit | |
| Authors |
|
| Supervisors |
|
| Language |
|
| Institution | University of Zurich |
| Faculty | Faculty of Economics, Business Administration and Information Technology |
| Number of Pages | 69 |
| Date | 2016 |
| Abstract Text | It is well-known that international stock portfolios greatly benefit from risk and return potentials in foreign markets, yet their constituents typically involve two risk factors: market risk and foreign exchange risk. We aim to uncouple these two immanent risks, but model them in parallel within a single model to foster a deeper understanding of the risk characteristics of international stock portfolios. To accommodate the empirical observation that almost all financial assets exhibit leptokurtosis, volatility clustering and mild asymmetry as well as joint extreme behaviour and possibly time-unstable dependence, we propose various extensions of the meta-elliptical t distribution of Fang, Fang and Kotz (2002), Fang, Fang and Kotz (2005) and Paolella and Polak (2015a). First, we allow for time variation by means of the copula parameters and second, we augment the structure of the copula function so that subgroups may assume different degrees of tail dependence. Furthermore, we propose a new calibration method of the conditional correlation models which greatly enhances the model performance, without causing any additional computational burden. The proposed models are then tested both in-sample and out-of-sample and indicate a superior fit when compared to the corresponding multivariate GARCH models. The model performance is then also illustrated in terms of practical applications, namely the predictive quality of Value-at-Risk forecasts, the conditional asset pricing of the portfolio constituents and one dynamic hedging exercise. In summary, the proposed models provide many new insights and the DCC model of Engle (2002) in conjunction with the copula parametrization yields the most appropriate distributional model, when compared to both static and regime switching alternatives. |
| Export |
BibTeX
|