Μέτρηση κινδύνου μέσω της θεωρίας ακραίων τιμών
Risk measurement using extreme value theory
Πολύζου, Στεργιανή Α.
SubjectExtreme value theory ; Διαχείριση κινδύνου -- Στατιστικές μέθοδοι ; Χρηματοοικονομική ανάλυση ; Κατανομή (Οικονομική θεωρία)
The evaluation of risk in financial science, apart from the well-known parametric meth¬ods can also be performed efficiently through Extreme Value Theory (EVT). This Master Thesis presents and applies (using specific financial series) two basic methods of EVT: Block Maxima method (BM) and Peaks over Threshold (POT) method. The work is divided into two parts, the theoretical (Chapters 1, 2, 3) and the practical part (Chapter 4). The first part presents theoretical results concerning (a) the Generalized Extreme Value and Generalized Pareto (GEV and GPD) distributions, and (b) the estima¬tors of their parameters. Furthermore two well known risk measures are analysed: the Value at Risk (VaR) and the Expected Shortfall (ES) through EVT, historical simulation and through the Variance-Covariance method. The second part consists of the study of the extreme changes of three specific financial series (from the New York Stock Ex- change-NYSE stock market) through the techniques presented in the first part. Finally, using the same methodology, the extreme increases or decreases of the values of ten specific shares of Greek stock market is investigated, before and after recent economic crisis (2009).