|
This thesis aims at improving the effectiveness of VaR models by combining Credibility Theories----the Limited Fluctuation Credibility Theory and Greatest Credibility Theory(the Bayes Measurement and the Buhlmann Model) in Actuarial Science with the Historical Simulation, the Parametric Methods, and the Extreme Value Theory. It develops from an introduction to the Classical VaR models (the Historical Simulation, the Parametric Methods, and the Monte-Carlo Simulation) and the extreme analysis of VaR models (the Stress Testing and the Extreme Value Theory); based on a brief review of the Credibility Theories, it runs to combining these credibility models into the classical and extreme VaR model systems. With a tentative conclusion of “Credibility VaR Model”, I finally points out the significance and also the defect of this thesis. | 
发表于 2010-3-14 09:50
| 