%0 Book %A Marcel Jäger %D 2015 %C Hamburg, Germany %I Anchor Academic Publishing %@ 9783954899753 %T Backtesting Expected Shortfall %B A GARCH-EVT-Copula Approach %U https://m.anchor-publishing.com/document/304757 %X Advanced research on coherent risk measures revealed some weaknesses in the widely used value-at-risk (VaR) model; hence, coherent risk measures, such as expected shortfall (ES) became increasingly popular. Consequently, a publication of the Basel Committee in 2012 suggested moving from VaR to ES as the new risk measure for the minimum amount of capital to cover potential loss. The backtesting of models has an important role in the Basel framework, because it proofs the predictability of risk models. While the backtesting for VaR is simple and established in most financial institutions, the backtesting for ES is unexplored and widely unknown. Moreover, the discovery in 2011 that ES is not an elicitable risk measure suggested that backtesting might not be possible. First, this thesis models financial portfolio returns using an approach combining GARCH models, extreme value theory (EVT) and copula functions. This approach is applied in the numerical computer language MATLAB to a sample portfolio of shares from Bayrische Motoren Werke AG, Deutsche Bank AG, Adidas AG and Siemens AG. The theory is based on the advanced research of statistical properties of financial returns, known as stylized facts. Finally, this approach is used to compare the backtesting results of VaR and ES. The results show that ES is backtestable but the basic idea of its backtesting procedure differs from that of VaR. The statistical tests for backtesting ES require the storage of more information and a large number of scenarios to compute the significance and power of the tests. This relates to some technical challenges for applying the backtest to ES. %K Backtesting Expected Shortfall, Value at Risk, Expected Shortfall, Backtesting, Marktpreisrisiko, Copula, Extreme value theory, GARCH, Basel, VaR, ES, Conditional Value at Risk %G English