Risk Measurement Using Time Varying Extreme Value Copulas

Abstract

This thesis aims to estimate reliable risk measures by considering the dependent and heavy-tailed characteristics of the different insurance risk branches. Thus, it is aimed to model the skewed and heavy-tailed data more accurately by an extreme value theory (EVT) model which combines the time series and copula models. This approach is also called dynamic EVT, due to two-piece modelling, which can tackle the specific needs of insurance data.